Measurement and prediction of antimicrobial resistance in bloodstream infections by ESKAPE and Escherichia coli pathogens

Accepted Manuscript Title: Measurement and prediction of antimicrobial resistance in bloodstream infections by ESKAPE and Escherichia coli pathogens Authors: Giuseppe Vittorio De Socio, Paola Rubbioni, Daniele Botta, Elio Cenci, Alessandra Belati, Riccardo Paggi, Maria Bruna Pasticci, Antonella Mencacci PII: DOI: Reference:

S2213-7165(19)30123-7 https://doi.org/10.1016/j.jgar.2019.05.013 JGAR 936

To appear in: Received date: Revised date: Accepted date:

4 February 2019 30 April 2019 9 May 2019

Please cite this article as: De Socio GV, Rubbioni P, Botta D, Cenci E, Belati A, Paggi R, Pasticci MB, Mencacci A, Measurement and prediction of antimicrobial resistance in bloodstream infections by ESKAPE and Escherichia coli pathogens, Journal of Global Antimicrobial Resistance (2019), https://doi.org/10.1016/j.jgar.2019.05.013 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Measurement and prediction of antimicrobial resistance in bloodstream infections by ESKAPE and Escherichia coli pathogens

Giuseppe Vittorio De Socio1*, Paola Rubbioni2, Daniele Botta1, Elio Cenci3,

IP T

Alessandra Belati3, Riccardo Paggi3, Maria Bruna Pasticci1, Antonella

1

SC R

Mencacci3.

Unit of Infectious Diseases, Department of Medicine, University of Perugia,

Department of Mathematics and Computer Science, University of Perugia,

N

2

U

Santa Maria della Misericordia Hospital, Perugia; Italy.

Medical Microbiology, Department of Medicine, University of Perugia,

M

3

A

Perugia, Italy

Corresponding author: Giuseppe V. L. De Socio, MD, PhD. Clinica di

PT

*

ED

Santa Maria della Misericordia Hospital, Perugia, Italy

CC E

Malattie Infettive, Azienda Ospedaliera di Perugia, Piazzale Menghini 1, 06129 Perugia, Italy, Phone : +39-075-5784321 Fax : +39-075-5784346 ; E-

A

mail: [email protected]

Author ORCID code: 0000-0001-8774-4843 Email of all the authors:

Giuseppe Vittorio De Socio; [email protected] Paola Rubbioni; [email protected] Daniele Botta; [email protected]

Riccardo Paggi; [email protected]

SC R

Alessandra Belati, [email protected]

IP T

Elio Cenci; [email protected]

Maria Bruna Pasticci; [email protected]

A

N

U

Antonella Mencacci; [email protected]

A

CC E

PT

ED

M

Highlights  A cumulative antimicrobial resistance index (ARI) from ESKAPE and Escherichia coli pathogens was proposed, as key outcome measure of antimicrobial stewardship (AS) programs and as a tool to predict antimicrobial resistance (AMR) trend.  The cumulative ESKAPEEc mean ARI increased significantly from 0.200±0.01, observed in 2014, to 0.276±0.02, in 2018.  A mathematical prediction model for antimicrobial resistance trend was obtained.  In the absence of a specific AS program, antimicrobial pan-resistance could be expected within the next 8-15 years.

2

ABSTRACT Objectives: The aim of the study was to evaluate a cumulative antimicrobial resistance index (ARI) as a possible key outcome measure of antimicrobial stewardship (AS) programs and as a tool to predict antimicrobial resistance

IP T

(AMR) trend. Methods: Antibiotic susceptibility for Enterococcus faecium, Staphylococcus

SC R

aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, Enterobacter species, and Escherichia coli (ESKAPEEc)

U

pathogens, recovered from blood cultures during a five-years period (2014-

N

2018), was analyzed to obtain a cumulative ARI. For each antibiotic tested a

M

A

score of 0, 0.5, or 1 was assigned for susceptibility, intermediate resistance, or resistance, respectively, and the ARI was calculated by dividing the sum of

ED

these scores by the number of antibiotics tested. Cumulative ARI of

PT

ESKAPEEc microorganisms were compared, and a mathematical prediction model for antimicrobial resistance trend was obtained.

CC E

Results: 1,858 ESKAPEEc isolates were included in the study. The cumulative ESKAPEEc mean ARI increased significantly from 0.200 0.01,

A

observed in 2014, to 0.276 0.02, in 2018 (p< 0.001). In multivariable regression analysis, factors significantly associated with ARI≥0.5 were Enterococcus faecium, Klebsiella pneumoniae, Pseudomonas aeruginosa and Acinetobacter baumannii infection (p<0.001), and infection occurred after the year 2014 (p<0.05). Based on the prediction model obtained, in the absence of 3

any interventional measure, a tendency to pan-resistance of the ESKAPEEc group could be expected in the next 8-15 years. Conclusions: ARI could be a useful tool to measure the impact of AS programs on antimicrobial resistance. The increasing incidence of AMR rates

A

CC E

PT

ED

M

A

N

U

SC R

IP T

among ESKAPEEc organisms underscores the needing for AS programs.

4

Abbreviation antimicrobial resistance (AMR) antimicrobial stewardship (AS) antimicrobial resistance index (ARI)

IP T

ESKAPE (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and

SC R

Enterobacter species)

ESKAPEEc (Enterococcus faecium, Staphylococcus aureus, Klebsiella

ED

bloodstream infections (BSIs)

M

Blood cultures (BC)

A

Laboratory Information System (LIS)

N

Enterobacter species plus Escherichia coli)

U

pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa,

multi drug resistant (MDR)

PT

infectious disease specialist (IDS)

CC E

European Committee on Antimicrobial Susceptibility Testing (EUCAST) standard deviation (SD)

A

standard error (SE) odds ratio (OR) confidence limits (CI)

Keywords: 5

bacterial

A

CC E

PT

ED

M

A

N

U

SC R

IP T

resistance

6

INTRODUCTION

The World Health Organization has identified antimicrobial resistance (AMR) among the top public health concern of the 21st century [1]. In 2017, ECDC

IP T

declared Italy as one of the member States with the higher level of resistance in Europe [2]. Indeed, a major driver of AMR has been considered the misuse

SC R

of antimicrobials in humans and animals [3], with antimicrobial use accelerating the development of resistance [4]. The challenge with

U

antimicrobial prescribing lies in the need to balance two conflicting goals: the

N

provision of adequate therapy to treat documented or presumed infection, and

A

the minimization of antimicrobial use to avoid the emergence of antimicrobial

M

resistance.

ED

Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, Enterobacter species,

PT

and Escherichia coli pathogens, defined as ESKAPEEc by De Angelis et al.

CC E

[5], are responsible for a substantial percentage of severe infections and of multi-drug-resistant organisms (MDRO) in the hospital [6]. During the last

A

five years, in our hospital ESKAPEEc ranged from 75.7% to 78.9% of total bacterial bloodstream infections. In the era of increasing AMR with limited availability of new and effective antibiotic agents, antimicrobial stewardship (AS) programs have emerged as a fundamental component of health care systems [7], being theoretically able to 7

avert trends in antimicrobial resistances, thus preserving future antimicrobial susceptibility. Indeed, the primary reason why antimicrobial stewardship is necessary is specifically the growth of antimicrobial resistance [8]. Accordingly, antimicrobial resistance should be the main metric of the impact

IP T

of an AS program. Susceptibility reporting tests have shown to be crucial for targeting

SC R

antimicrobial therapy in single clinical cases. However, information on a single pathogen or a single antibiotic cannot be sufficient to measure the

U

efficacy of institutional programs of antibiotics use. Unfortunately, there are

N

no proposed aggregate measures of ASP success in the literature, considering

A

resistance to multiple antimicrobials in multiple organisms [9]. The low

ED

measuring its outcome.

M

evidence for the impact of AS on AMR is likely related also to difficulties in

The primary aim of this study was to identify a simple way to decode and

PT

predict AMR trend in the next years, based on surveillance of ESKAPEEc

CC E

organisms recovered from bloodstream infections. A cumulative ARI, obtained in the setting of a real-life, was generated to quantify hospital

A

antimicrobial resistance trend, and to be used by healthcare stakeholders to address specific AS programs. As secondary objectives evaluate factors associated with high ARI.

8

METHODS This was a monocentric observational retrospective study. Data collection and description Susceptibility testing results related to ESKAPEEc organisms, isolated from

IP T

blood cultures from patients admitted to the Perugia General Hospital from 1 January 2014 to 15 October 2018 were collected from the Laboratory

SC R

Information System (LIS) archive (TD-Synergy, Siemens). Identification of ESKAPEEc organisms in blood samples was considered a proxy of really

U

bloodstream infections (BSI), given its high positive predictive value.

N

Pathogens different from ESKAPEEc organisms were excluded from the

A

analysis. For each pathogen/patient combination, only the first BSI episode

M

was considered.

ED

For years 2016, 2017, and 2018, BSI were classified as hospital-onset or community-onset if the first positive blood culture had been collected >48 h

PT

after or ≤48 h before admission, respectively. The above differentiation for

CC E

years 2014 and 2015 was not done, as the day of hospital admission, for these years, was not recorded in the LIS archive.

A

The Microbiology Laboratory provides diagnostic services to the 800-bed General Hospital of Perugia, Italy, serving a population of around 200.000 people. In our hospital an infectious disease specialist (IDS) is available on call basis in all hospital wards at bedside 24h/7 days. During the period 2014-2018 9

there wasn’t local restriction policy of antibiotic use, including carbapenems or other broad-spectrum antibiotics, nor a specific AS program with preauthorization requirements for antimicrobial agents [10]. Only an “educational/persuasive” approach was used by the IDS consultant, only upon

IP T

request by patient ward’s clinicians. To date, no change in the AS policy has been made in the hospital. However,

SC R

in the last observational year (2018), the laboratory management of blood

cultures has improved. First, thanks to a satellite incubator, all blood cultures

U

are rapidly incubated within 1 h from collection, also outside of the operating

N

hours of the Microbiology laboratory. Second, after the introduction of a

A

laboratory automated system, a new method to process positive blood cultures

M

has been implemented with 8-h digital reading of subculture plates, followed

ED

by immediate identification and antimicrobial susceptibility testing. This method has been shown to significantly reduce time-to-report and to shorten

PT

the duration of broad-spectrum empiric therapy of about 32 h [11]. Moreover,

CC E

from 2018, a more stringent collaboration between the ID specialist and clinical microbiologist has been established, with an everyday briefing to

A

discuss the most serious cases and blood culture results from surgical and intensive care unit. Blood culture and susceptibility testing Blood cultures were received within one hour from collection from different hospital wards inoculated into BD BACTEC Plus Aerobic/F and BD 10

BACTEC Lytic/10 Anaerobic/F bottles and incubated immediately in the BD BACTEC FX instrument (Becton Dickinson). Positive BC were processed for Gram staining and subculture on solid media, performed manually or, in the last two years of the study, automatically on Work Cell Automation System

IP T

(Becton Dickinson), as previously described [11]. Colonies were identified using the Bruker MALDI Biotype instrument (Bruker Daltonik GmbH,

SC R

Bremen, Germany), as described elsewhere [12]. Antimicrobial susceptibility testing was performed with the BD Phoenix automatic system (Becton

U

Dickinson). For methicillin-resistance screening, cefoxitin disk diffusion test

N

was performed on S. aureus, while the combination disk test was used for

A

ESBL detection in Enterobacteriaceae. Klebsiella pneumoniae, A. baumannii,

M

or P. aeruginosa organisms, suspected to be multi-drug-resistant (e.g., patient

ED

surveillance swabs or other specimens positive for multi-resistant K. pneumoniae, A. baumanni or P. aeruginosa and/or colonies grown on blood

PT

CNA agar plates) were tested with Sensititre microdilution method (Thermo

CC E

Fisher Scientific, Cleveland, OH, USA). Klebsiella isolates were tested for carbapenemases by the Xpert Carba-R assay (Cepheid, Sunnyvale, CA,

A

USA), according to manufacturer’s instructions. MIC values were translated in clinical categories (S/I/R) according to current EUCAST criteria [13-17]. Acinetobacter baumannii, or P. aeruginosa organisms were considered multi drug resistant (MDR) if exhibited resistance to at least one antibiotic in at least three relevant antimicrobial categories [18]. 11

Antimicrobial Resistance Index and prediction model for Antimicrobial Resistance trend To calculate ARI the model for measure antibiotic resistance in P. aeruginosa

IP T

in patients with cystic fibrosis, used by Ewing et al. [19] was followed. Briefly, for each antibiotic tested in each ESKAPEEc microorganism, a score

SC R

of 0 for susceptibility, 0.5 for intermediate resistance, or 1 for resistance were assigned, and the ARI was calculated by dividing the sum of these scores by

U

the number of antibiotics tested, giving a maximum score of 1. Thus, ARI=0

N

corresponded to a pan-susceptible organism, and ARI=1 to a pan-resistant

A

one.

M

Then, the ARI of each ESKAPEEc species and the cumulative ARI for the

ED

entire ESKAPEEc group were obtained for each year included in the study, and were compared.

PT

A low-intermediate antimicrobial resistance was defined whether cumulative

CC E

ARI was <0.5; high-very high resistance was for ARI≥0.5. Factors associated with cumulative ARI≥0.5 were evaluated.

A

The prediction of ARI in the overcoming years was obtained from the data of the cumulative ARI for the ESKAPEEc organisms in each observational year, as follows: defining X_n the ARI in the year “n”, the annual ARI increase (i.e. the increase of ARI between the two consecutive years “n-1” and “n”) was 12

I_n= X_n-X_(n-1). Hence, the rate T_n of the year-by-year increase/decrease was given by the formula T_n=I_n/X_(n-1).

T=(T_1+T_2+…+T_n)/n.

SC R

considered the average T of the year-by-year T_n rates

IP T

Now, in order to use a T_n rate which did not depend on the year “n”, we

This value T, considering the current conditions stable, led us to predict ARI

U

in next years, according to the following formula:

N

X_n=(1+T)X_(n-1), for n>1.

A

Ethics statement

M

Not required. Samples were collected and results were delivered to wards as

ED

part of standard care. Data included in the database were extracted from the archive of microbiology laboratory and de-identified before access. No

PT

personal information was stored in the study database. All collected data were

CC E

analyzed anonymously.

A

Statistical analysis Standard descriptive statistics were used to summarize data, such as mean, standard deviation (SD) or standard error (SE), and percentage, as appropriate. Differences for categorical variables were evaluated by Pearson's 2

test. Bacterial isolates were compared each one vs any others. Unadjusted 13

absolute changes in the ARI was calculated from the 2015 to 2018, at each time point by ANOVA test. Mantel-Haenszel test was used for trend analysis. The univariable and multivariable logistic regression analyses were used to evaluate factors associated with ARI≥0.5. The dependent variable was

IP T

ARI≥0.5 as a dichotomic variable in the logistic models. Variables with significance level <0.1 were included in multivariable analysis. The results of

SC R

the logistic analysis are presented as odds ratio (OR) and 95% confidence limits (CI). Multivariable analysis included the following variables: age,

U

hospital ward, bacterial isolates, observational year. The analysis was not

N

adjusted for hospital-onset BSIs, as data for year 2014 and 2015 were not

A

available.

A

CC E

PT

ED

22.0 (SPSS Inc, Chicago, Ill).

M

All statistical tests were performed using SPSS statistical package, release

14

RESULTS ARI in Perugia General Hospital from 2014 to 2018 A total of 8,348 consecutive positive blood culture were analysed in the five years study period (2014-2018). After the exclusion of 6,490 samples

IP T

according to the exclusion criteria (4,640 not ESKAPEEc microorganisms and 1,850 redundant samples), 1,858 ESKAPEEc isolates were included in

SC R

the study: 385 in year 2014, 415 in 2015, 391 in 2016, 408 in 2017 and 259 in 2018 (until 15 October 2018).

U

Patients’ demographic characteristics and bacterial isolates included in the

N

study are depicted in Table 1. No statistical difference was observed for

A

patients age, gender and hospital wards. More than the half of BSIs were

M

caused by two bacterial species: E. coli, the most frequent pathogen (34.0%),

ED

and S. aureus (25.5%), with no significant differences among the 5 years. Selected bacterial species were not evenly distributed in the five observational

PT

years. Hospital-onset BSIs increased significantly from 37.4% in 2016 to

CC E

50.4% in 2018 (p=0.002). A significant increase of ARI was observed for some, but not all, bacterial

A

species (Table 2). The ESKAPEEc group cumulative ARI progressively and significantly increased during the five-years study period from 0.200±0.01 to 0.276±0.02 (p<0.001) (Figure 1A), due to the increase of isolates with high ARI ≥0.5. (Figure 1B).

15

ARI in hospital-onset BSI was significantly (p<0.001) higher than in community-onset infection (Figure 2). In multivariable regression analysis (Table 3) including age, gender, hospital ward, ESKAPEEc organism and observational year, factors significantly

IP T

associated with ARI≥0.5 were: E. faecium (OR 15.052; CI 95% 8.07528.059), K. pneumoniae (OR 52.891; CI 95% 30.531-91.626), P. aeruginosa

SC R

(OR 4.889; CI 95% 2.369-10.088), A. baumannii infection (OR 296.811; CI

95% 115.696-761.451) (p<0.001) and infection occurred after the 2014 year

U

(p<0.05). An increased OR was observed both at uni- and multivariable

A

N

analysis, in any subsequent observational year following 2014.

M

Prediction Model of Antimicrobial Resistance

ED

Since from 2014 (year 0) to 2018 (year 4), the ARI data were respectively X_0=0.200; X_1=0.221; X_2=0.244; X_3=0.284; X_4=0.276,

PT

than, according to the formula

CC E

I_n= X_n-X_(n-1),

the annual increase/decrease of ARI for each observational year was:

A

I_1= 0.021; I_2=0.023; I_3= 0.04; I_4= -0.008. Next, according to the formula T_n=I_n/X_(n-1), the rate T_n of each year-by-year increase/decrease was T_1=0,105; T_2=0,104; T_3=0,164; T_4= -0,028. 16

The average T from 2014 to 2018 was T=(T_1+T_2+T_3+T_4)/4=0,086, hence the corresponding annual prediction given by the formula X_n=(1+T)X_(n-1) provides X_n = 1,086 X_(n-1)

IP T

(Figure 3, line A). Because the I_4, for the year 2018, showed an opposite trend with respect to the previous years, the average T from 2014 to 2017 was

SC R

calculated as well, resulting in T=(T_1+T_2+T_3)/3 = 0,124, leading to the alternative annual prediction

U

X_n = 1,124 X_(n-1)

A

CC E

PT

ED

M

A

N

(Figure 3, line B).

17

DISCUSSION The advent of multidrug resistance among pathogenic bacteria is imperilling the worth of antibiotics, which have previously transformed medical sciences. The global spreading of AR, due predominantly to the

IP T

misuse of these agents and unavailability of new drugs [20] points for the importance of AS programs aimed to optimize antibiotic use and minimize

SC R

resistance [21].

There is a growing interest in the use of laboratory or clinical records to

U

provide information supporting AS programs and measure their impact in

N

clinical practice. Although it is widely accepted that AS is effective in

A

increasing compliance with antibiotic policy and reducing duration of

M

antibiotic treatment, nevertheless there is a surprising paucity of literature

ED

demonstrating the real impact of AS on antimicrobial resistance [22], probably due to the difficulty in measuring this outcome [23]. It has been

PT

reported that decreased use of antibiotics probably does not increase mortality

CC E

and possibly reduces length of hospital stay [24], and reduction in Clostridium difficile infections can be considered a measure of outcome of AS

A

intervention [ 10]. This study proposes a simple method measuring AR in a hospital and

predicting AR trend, based on cumulative ARI of all the ESKAPEEc organisms. According to this model, we found that antimicrobial resistance patterns in blood samples isolates progressively increased from 2014 to 2018 18

in Perugia General Hospital, and, in the absence of a specific AS program, pan-resistance could be expected within the next 8-15 years. We found that ARI of some bacterial species did not significantly vary in different years (Table 2), while the cumulative ARI of the entire

IP T

ESKAPEEc group significantly increased from 2014 to 2018 (Figure 1). This finding can be explained by the fact that cumulative ARI does not depend

SC R

only on the resistance of single bacterial species, but also on the relative percentage of highly resistant organisms isolated each year. This result

U

highlights the usefulness of ARI in depicting the real weight of AR in a

N

particular setting. The bacterial species mainly contributing to high/very high

A

ARI in multivariable analysis were E. faecium, K. pneumoniae, A.baumannii

M

and P. aeruginosa and, as expected, ARI was higher in hospital-onset

ED

compared with community-onset infections, in line with the notion that infections by MDR organisms are more common in hospital than in

PT

community. Similarly, the progressive increase of AR in our Hospital, with

CC E

pan-resistance predicted by 8-15 years is dramatically in line with the global AR crisis, rendering infections virtually untreatable in the year 2050 [25].

A

Actually, in our setting, a minimal reversal AMR trend was observed in

2018. Although we cannot exclude that this finding is incidental, it could nevertheless be a consequence of two changes in the management of sepsis in our hospital in the last year: a significant reduction in time to report for BC results, after the introduction of molecular technologies and automation in the 19

laboratory work-flow on positive BC [11], and a fruitful, close collaboration between the ID and clinical microbiologist. Thus, assuming that the reversal trend in AMR in 2018 was due to the improved management of sepsis in our Hospital, pan-resistance can be predicted by 15 years. On the other hand,

IP T

according to a less optimistic vision, considering the decreased 2018 ARI as a coincidence thus excluding it from the formula, pan-resistance could be

SC R

expected within 8 years. Nevertheless, according the prediction model

proposed in this study, an increasing trend in AMR is expected in the next

U

years in our Hospital.

N

ARI can be intelligible to non-experts and useful to experts, and could be

A

applied as a measure tool of different AS programs over time. It is not clear

M

what kind of AS program could be more effective in our context to contain

ED

AMR trends. Many different methods to improve prescription of antibiotics in hospitals have been studied. These programs can be distinguished in

PT

persuasive or restrictive methods to reduce unnecessary antibiotic use.

CC E

Persuasive methods advised physicians about how to prescribe or gave them feedback about how they prescribed. Restrictive methods put a limit on how

A

they prescribed, and are considered able to obtain a faster effect than persuasive methods [10]. A recent meta-analysis supports the use of restrictive interventions when the need is urgent, but suggests that persuasive and restrictive interventions are equally effective after six months from their application [10]. However, the time frame of 6 months may be too short to 20

produce significant results, in terms of both reduction of incidence of infections by MDR organisms and of ARI. Moreover, ARI could be not appropriate to evaluate the efficacy of an AS program in time period shorter than one year. On the contrary, in the clinical practice, a short-term tool

IP T

would be useful. We found that 2018 ARI, although lower than 2017, was higher than in

SC R

previous years. It is also possible that the lack of a specific implementation of AS policy in our hospital could explain the increase of ARI observed in 2018

U

year. Furthermore, our “soft persuasive” strategy on antimicrobial

N

prescription may not have succeeded in reducing unnecessary use of

A

antibiotics and the widespread prescription of broad-spectrum antimicrobials.

M

On the contrary, this study underlines the needing of urgent strong specific

ED

actions in the antimicrobial and infection control, according to recent recommendations from the European Centre for Disease Prevention and

PT

Control [2] and the Italian Group for Antimicrobial Stewardship (GISA) [26].

CC E

The present study has some strengths and limitations that deserve some comments. ARI, a tool to get a view of AMR trend simple to be calculated,

A

was set up based on the most appropriate an up-to date methods for antimicrobial susceptibility testing, that included all first isolates of ESKAPEEc organisms’ group per single patient over 5 years. As this was a single-centre study, performed in a teaching Hospital with a medically complex patient population, replication of our results in other 21

Institutions and in other patient populations is necessary to enhance the generalizability of our findings. In addition, we are not able to evaluate in multivariate analysis the role of hospital-onset BSIs as data not availability for year 2014 and 2015 in database.

IP T

To minimize the bias due to the fact that, for a few antibiotic molecules, data on susceptibility of single bacterial species were not always

SC R

available throughout the 5-years study period, due to availability of new drugs or laboratory diagnostic tests, the ARI was normalized by the number of

U

molecules tested. Other potential limitations include the absence of any

N

consideration in the prediction model of infection-control interventions and

A

on the difference among antibiotics and single species as the development of

M

resistance to different antibiotics is considered equally, independent of the

ED

class and the bacterial species.

In conclusion, the increasing ARI rates among ESKAPEEc organisms

PT

in our Hospital underscores the needing for a systematic approach and AS

CC E

programs to effectively manage AMR in BSI. The method proposed, in combination with others measure tools of AS, could be useful to measure the

A

effects of such programs in our hospital, but also to compare them with those obtained with different approaches in other settings.

22

Declarations Author’s contributions Study concept and design: GVDS. Statistical expertise: GVDS. Drafting of the manuscript: GVDS, PR, AM. Mathematical expertise and mathematical

IP T

model PR. Analysis, interpretation, discussion of data and critical revision of

SC R

the manuscript for important intellectual content: all the authors. The authors had full access to the data and take responsibility for their integrity. All authors have read and agreed to the final manuscript as written.

A

N

clinical practice (National Health Service).

U

Funding: No has been provided by any private or public actor beyond current

M

Ethical Approval: The study follows the principles of the Declaration of Helsinki. Data included in the database were extracted from the archive of

ED

microbiology laboratory and de-identified before access. No personal

PT

information was stored in the study database. All collected data were analyzed anonymously.

A

CC E

Competing Interests: The authors declared no conflict of interest.

23

24

A ED

PT

CC E

IP T

SC R

U

N

A

M

References 1.

World Health Organization (WHO). Antimicrobial Resistance— Global Report on Surveillance, Geneva:WHO (2014). Available at http://apps.who.int/iris/bitstream/handle/10665/112642/978924156474

IP T

8_eng.pdf;jsessionid=C52D86186224786B74FB7052D5492AFE?sequ ence=1 Accessed December 6 2018

SC R

2. European Centre for Disease Prevention and Control. ECDC (2017)

country visit to Italy to discuss antimicrobial resistance issues.

U

Stockholm: ECDC; Available at

N

https://ecdc.europa.eu/sites/portal/files/documents/AMR-country-visit-

A

Italy.pdf Accessed December 6 2018

M

3. McIntosh W, Dean W. Factors associated with the inappropriate use of

ED

antimicrobials. Zoonoses Public Health 2015;62 Suppl 1:22-8. doi: 10.1111/zph.12169. Epub 2014 Dec 2. Review.

PT

4. Wise R. BSAC Working Party on The Urgent Need: Regenerating

CC E

Antibacterial Drug Discovery and Development. J Antimicrob Chemother. 2011;66(9):1939-40. doi: 10.1093/jac/dkr261.

A

5. De Angelis G, Fiori B, Menchinelli G, D'Inzeo T, Liotti FM,

Morandotti GA, Sanguinetti M, Posteraro B, Spanu T. Incidence and antimicrobial resistance trends in bloodstream infections caused by ESKAPE and Escherichia coli at a large teaching hospital in Rome, a 9-

25

year analysis (2007-2015). Eur J Clin Microbiol Infect Dis. 2018 37(9):1627-1636. doi: 10.1007/s10096-018-3292-9. 6. Rice LB. Progress and challenges in implementing the research on

ESKAPE pathogens. Infect Control Hosp Epidemiol. 2010;31 Suppl

IP T

1:S7-10. doi: 10.1086/655995. 7. Taylor G. The importance of stewardship. Can Commun Dis Rep.

SC R

2015;18;41(Suppl 4):3-4. eCollection.

8. Bartlett JG. A call to arms: the imperative for antimicrobial

U

stewardship. Clin Infect Dis 2011;53(suppl 1):S4–S7. doi:

N

10.1093/cid/cir362.

A

9. Schulz LT, Fox BC, Polk RE. Can the antibiogram be used to assess

M

microbiologic outcomes after antimicrobial stewardship interventions?

ED

A critical review of the literature. Pharmacotherapy. 2012;32(8):668– 76.

PT

10. Davey P, Brown E, Charani E, Fenelon L, Gould IM, Holmes A,

CC E

Ramsay CR, Wiffen PJ, Wilcox M. Interventions to improve antibiotic prescribing practices for hospital inpatients. Cochrane Database Syst

A

Rev. 2013;30;(4):CD003543. doi: 10.1002/14651858.CD003543

11. De Socio GV, Di Donato F, Paggi R, Gabrielli C, Belati A, Rizza G,

Savoia M, Repetto A, Cenci E, Mencacci A. Laboratory automation reduces time to report of positive blood cultures and improves

26

management of patients with bloodstream infection. Eur J Clin Microbiol Infect Dis. 2018; doi: 10.1007/s10096-018-3377-5. 12. Leli C, Cenci E, Cardaccia A, Moretti A, D’Alò F, Pagliochini R,

Barcaccia M, Farinelli S, Vento S, Bistoni F, Mencacci A Rapid

IP T

identification of bacterial and fungal pathogens from positive blood cultures by MALDI-TOF MS. Int J Med Microbiol 2013; 303(4):205-

SC R

209.

13. European Committee on Antimicrobial Susceptibility Testing, Version

U

4.0 (2014) availabe at

N

http://www.eucast.org/fileadmin/src/media/PDFs/EUCAST_files/Break

A

point_tables/Breakpoint_table_v_4.0.pdf Accessed December 6 2018

ED

5.0 (2015) available at

M

14. European Committee on Antimicrobial Susceptibility Testing, Version

http://www.eucast.org/fileadmin/src/media/PDFs/EUCAST_files/Break

CC E

2018

PT

point_tables/v_5.0_Breakpoint_Table_01.pdf Accessed December 6

15. European Committee on Antimicrobial Susceptibility Testing, Version

A

6.0 (2016) available at http://www.eucast.org/fileadmin/src/media/PDFs/EUCAST_files/Break point_tables/v_6.0_Breakpoint_table.pdf Accessed December 6 2018

16. European Committee on Antimicrobial Susceptibility Testing, Version

7.1 (2017) available at 27

http://www.eucast.org/fileadmin/src/media/PDFs/EUCAST_files/Break point_tables/v_7.1_Breakpoint_Tables.pdf Accessed December 6 2018 17. European Committee on Antimicrobial Susceptibility Testing, Version

8.0 (2018) available at

IP T

http://www.eucast.org/fileadmin/src/media/PDFs/EUCAST_files/Break point_tables/v_8.0_Breakpoint_Tables.pdf Accessed December 6 2018

SC R

18. Magiorakos AP, Srinivasan A, Carey RB, Carmeli Y, Falagas ME,

Giske CG, Harbarth S, Hindler JF, Kahlmeter G, Olsson-Liljequist B,

U

Paterson DL, Rice LB, Stelling J, Struelens MJ, Vatopoulos A, Weber

N

JT, Monnet DL. Multidrug-resistant, extensively drug-resistant and

A

pandrug-resistant bacteria: an international expert proposal for interim

M

standard definitions for acquired resistance. Clin Microbiol Infect.

ED

2012;18(3):268-81. doi: 10.1111/j.1469-0691.2011.03570.x 19. Ewing J, McCaughan J Moore J, Fairley D, Sutherland B, Reid A,

PT

Downey D Relative resistance index (RRI) - a scoring system for

CC E

antibiotic resistance in Pseudomonas aeruginosa.- Br J Biomed Sci. 2017;74(4):198-202. doi:10.1080/09674845.2017.1338500.

A

20. Aslam B, Wang W, Arshad MI, Khurshid M, Muzammil S, Rasool

MH, Nisar MA, Alvi RF, Aslam MA, Qamar MU, Salamat MKF, Baloch Z. Antibiotic resistance: a rundown of a global crisis. Infect Drug Resist. 2018;11:1645-1658. doi: 10.2147/IDR.S173867.

28

21. Cunha CB, Opal SM. Antibiotic Stewardship: Strategies to Minimize

Antibiotic Resistance While Maximizing Antibiotic Effectiveness. Med Clin North Am. 2018;102(5):831-843. doi: 10.1016/j.mcna.2018.04.006

IP T

22. Schulz LT, Fox BC, Polk RE. Can the antibiogram be used to assess

microbiologic outcomes after antimicrobial stewardship interventions?

76. doi: 10.1002/j.1875-9114.2012.01163.x.

SC R

A critical review of the literature. Pharmacotherapy. 2012;32(8):668-

U

23. Tamma PD, Cosgrove SE. Antimicrobial stewardship. Infect Dis Clin

N

North Am. 2011;25(1):245-60. doi: 10.1016/j.idc.2010.11.011.

A

24. Davey P, Marwick CA, Scott CL, Charani E, McNeil K, Brown E,

M

Gould IM, Ramsay CR, Michie S. Interventions to improve antibiotic

ED

prescribing practices for hospital inpatients. Cochrane Database Syst Rev. 2017;2:CD003543. doi: 10.1002/14651858.CD003543.pub4

PT

25. The Review on Antimicrobial Resistance, Chaired by Jim O’Neill.

CC E

2014; Available at https://amrreview.org/sites/default/files/AMR%20Review%20Paper%20-

A

%20Tackling%20a%20crisis%20for%20the%20health%20and%20wea lth%20of%20nations_1.pdf Accessed December 6 2018.

26. Menichetti F, Falcone M, Lopalco P, Tascini C, Pan A, Busani L,

Viaggi B, Rossolini GM, Arena F, Novelli A, De Rosa F, Iannazzo S, Cohen J; for GISA (Italian Group for Antimicrobial Stewardship). The 29

GISA call to action for the appropriate use of antimicrobials and the control of antimicrobial resistance in Italy. Int J Antimicrob Agents.

A

CC E

PT

ED

M

A

N

U

SC R

IP T

2018 ;52(2):127-134. DOI: 10.1016/j.ijantimicag.2018.05.010.

30

Figure Legends. Figure 1 A) Mean of antimicrobial resistance index (ARI) of all the ESKAPEEc organisms in year 2014-2018.

IP T

B) Whole grey bar: percentage of isolates with a low/intermediate resistance (ARI<0.5); Striped bar: percentage of isolates with high/very high

ED

M

A

N

U

SC R

resistance (ARI≥0.5).

PT

Figure 2

CC E

Antimicrobial resistance index (ARI) from 2016 to 2018 in hospital-onset or

A

community-onset blood stream infections (BSIs).

31

IP T SC R U

Figure 3

N

Prediction model of antimicrobial resistance index (ARI) for the ESKAPEEc.

M

A

A) Prediction including data from 2018 (whole line, A)

A

CC E

PT

ED

B) Prediction excluding data from 2018 (dashed line, B).

32

I N U SC R

TABLES Table 1.

A

Patients’ demographic characteristics and bacterial isolates from positive blood cultures in observational five-years study

All 1858 68.0 1077 ±19.1 (58.0)

2014 385 67.7 223 ±18.8 (57.9)

2015 415 67.5 230 ±19.3 (55.4)

2016 391 67.7 234 ±20.6 (59.8)

2017 408 69.2 239 ±18.4 (58.6)

2018 259 66.2 151 (58.3) ±20.7

1282 302 (69.0) 274 (16.3) (14.7) 426 577 (42.5) (57.5) 603 473 (32.5) 130 (7.0) (25.5) 1255 631 131 (7.1) (67.5) (34.0) 315 61 (3.3) (17.0)

287(74.5) 48 (12.5) 50 (13.0)

271 67(16.1) (65.3) 77(18.6)

276 63(16.1) (70.6) 52 (13.3)

279 71 (17.4) (70.8) 48 (11.8)

159(61.4) 53 (20.5) 47 (18.1)

NA NA a 132 103 (34.3) 29 (7.5) (26.8) 253 147 28 (7.3) (65.7) (38.2) 52 (13.5) 12 (3.1)

NA NA

130 218 (37.4) (62.6) 137 127 113 100 (33.0) (32.5) 24 (5.8) 27(6.9) (27.2) (25.6) 278 264 147(35.4) 129 27 (6.5) 33 (8.4) (67.0) (67.5) (33.0) 58 (14.0) 70 (17.9) 12 (2.9) 5 (1.3)

168 233 (41.9) (58.1) 121(29.7) 87 (21.3) 34 (8.3) 287 132 28 (6.9) (70.3) (32.4) 88 (21.6) 22 (5.4)

128 (50.4) 126 (49.6)

0.002

86 (33.2) 70 (27.0) 16 (6.2) 173 (66.8) 76 (29.3) 15 (5.8) 47 (18.1) 10 (3.9)

0.704 0.283 0.644 0.704 0.161 0.733 0.014 0.025

TE

D

Variable Patients Age (years; mean, SD) orooorganisms Men

M

period.

CC EP

Hospital ward Medicine Surgical Intensive Care

A

Hospital-onset BSIs Community-onset BSIs Bacterial isolates Gram-positive bacteria S. aureus E. faecium Gram-negative bacteria E. coli P. aeruginosa K. pneumoniae A. baumannii

p 0.346 0.783 0.144*

I N U SC R

A

CC EP

TE

D

M

A

Enterobacter spp. 117 (6.3) 14 (4.6) 34 (8.2) 27 (6.9) 17 (4.2) 25 (9.7) 0.004 Resistant isolates Methicillin-R S. aureus 132(27.9) 35 (34.0) 32 (28.3) 26(26.0) 21 (24.1) 18 (25.7) 0.581 Vancomycin-R E. 33 (25.4) 4 (13.8) 6 (25.0) 8 (29.6) 9 (26.5) 6 (37.5) 0.466 E. coli R to 3dt gen 152 21 (35) 42 (30.0) 35 (27.6) 31(25.4) 23 (35.4) 0.529 faecium P. aeruginosa MDR 11(8.4) 4(14.3) 3(11.1) 2 (6.1) 2(7.1) 0 (0.0) 0.527 cephems (29.6) K. pneumoniae R to 123 19 (36.5) 28(48.3) 30 (42.9) 21 (23.9) 25 (53.2) 0.004 A. baumannii MDR 54 (88.5) 10 (83.3) 8 (66.7) 5 (100) 21 (95.5) 10 (100) 0.062 carbapenems (39.0) Enterobacter spp R to 33 (28.2) 3 (21.4) 14 (41.2) 0 (0.0) 7 (41.7) 9 (36.0) 0.003 3 cephems Values are Number (%) unless otherwise specified.; p = ANOVA test for multiple comparison of continuous variable, Chisquared test for categorical variables; *Mantel-Haenszel test. All p-values are referred to the comparison between the individuals with or without the selected bacteria: e.g. those with versus those without the isolate. MDR, multi drug resistant. NA, data not available. Missing data for Hospital or community acquired BSIs definition: 385 in 2014, 415 in 2015, 43 in 2016, 7 in 2017, and 5 in 2018.

34

I N U SC R

Table 2.

Antimicrobial resistance index (ARI) in any observational years in ESKAPEEc bacteria. 2017

2018

D

TE

CC EP A

P. aeruginosa

Enterobacter spp

1 0.5 0

0.1680.01 0.1500.01 0.1690.01 0.1620.01 0.1600.01 0.776

A. baumannii

Trend

p*

0.3390.03 0.3540.04 0.4200.03 0.4160.03 0.4530.05 0.184

S. aureus

K. pneumoniae

2016

M

E. faecium

2015

A

2014

1 0.5 0

0.3350.05 0.5440.05 0.5160.04 0.4870.04 0.5570.04 0.011

1 0.5 0

0.6060.07 0.6340.07 0.8660.01 0.8740.03 0.8500.02 <0.001

1 0.5 0

0.2100.05 0.1260.04 0.1690.04 0.1380.04 0.0960.04 0.568

1 0.5 0

0.1290.03 0.1880.02 0.1230.01 0.2410.04 0.1710.03 0.056

1 0.5 0

35

I N U SC R

E. coli

0.1210.01 0.1190.01 0.1370.01 0.1340.01 0.1670.02 0.203

0

0.2000.01 0.2210.01 0.2440.01 0.2840.01 0.2760.02 <0.001

A

Cumulative ESKAPEEc

1 0.5

1 0.5 0

A

CC EP

TE

D

M

Values are Mean  Standard Error. *p= ANOVA for variance.

36

I N U SC R

Table 3

Univariable and multivariable analysis for factors associated with high-very high ARI (ARI ≥0.5). ARI ≥0.5 *OR and 95% CI 315

68.0 781 ±19.1 1077 (42.0) (58.0) 1282 302 (69.0) 274 (16.3) (14.7) 577 426 (57.5) (42.5)

66.2 121(38.4) ±16.9 194 (61.6) 202 39 (12.4) (64.1) 74(23.5)

A

CC EP

Community-onset Hospital-onset BSIs BSIs Bacterial isolates E. coli S. aureus E. faecium P. aeruginosa. K. pneumoniae A. baumannii Enterobacter spp. Observational 2014 year 2015

A

All 1858

M

0.994 (0.988-1.00) 1 (ref) 1.198 (0.9351.536) 1 (ref) 0.793 (0.5491.978 1.146)(1.4572.687) 92 (43.8) 1 (ref) 118(56.2) 2.023 (1.4882.752)

D

TE

Variable Patients orooorganisms Age (years; Women mean, SD) Men Hospital ward Medicine Surgical Intensive Care

631 473 (34.0) 130 (25.5) 131 (7.0) 315 (7.1) 61 (3.3) (17.0) 117 (6.3)

17(5.4) 1 (0.3) 37 (11.7) 16 (5.1) 186 54 (17.1) (59.0) 4 (1.3)

385 415

39 (10.1) 57 (13.7)

1 (ref) 0.077 (0.01-0.577) 14.369 (7.7745.025 (2.46826.560) 52.077 10.233)(30.602278.622 88.621) (110.6911.29(0.422-3.870) 701.325)

** OR and 95% CI

p* 0.047

** p NS

0.153 0.041 <0.001

NS NS

0.013 <0.001 <0.001 <0.001 <0.001 0.664

0.072 (0.01015.052 0.544) (8.0754.889 (2.36928.059) 52.891 10.088)(30.531296.811 91.626) (115.6961.060(0.342-3.283) 761.451)

0.011 <0.001 <0.001 <0.001 <0.001 0.919

1.819 (1.0303.208)

0.039

<0.001§ 1 (ref) 1.413 (0.9162.179) 37

I N U SC R

A

2016 391 65 (16.6) 1.769 (1.1572.357 (1.3572017 408 95 (23.3) 2.693 (1.8002.459 2.705) 4.095)(1.4512018 259 59 (22.8) 2.617 4.167 4.029)(1.6854.167)(2.270Values are Number (%). CI = Confidence Interval. OR = Odds Ratio. NS= not significant. 4.065) 7.651) § Mantel-Haenszel test for trend among observational years. *Univariable analysis; ** Multivariable analysis.

0.002 0.001 <0.001

A

CC EP

TE

D

M

**Multivariable analysis included the following variables: Age, hospital ward, bacterial isolates, observational years. The analysis was not adjusted for hospital onset BSIs as data not availability for year 2014 and 2015.

38