Point prevalence survey of antimicrobial use and healthcare-associated infections in Belgian acute care hospitals: results of the Global-PPS and ECDC-PPS 2017

Background The point prevalence survey of healthcare-associated infections (HAIs) and antimicrobial use organized by the European Centre for Disease Prevention and Control (ECDC-PPS) and the Global Point Prevalence Survey of antimicrobial consumption (Global-PPS) were simultaneously performed in Belgian acute care hospitals in 2017. Methods Belgian acute care hospitals were invited to participate in either the ECDC or Global-PPS. Hospital/ward/patient-level data were collected between September–December 2017. All patients present in the wards at 8 a.m. on the day of the PPS were included. The data of the ECDC and Global-PPS on antimicrobial consumption were pooled. Detailed data on HAIs were analysed for ECDC-PPS. Results Overall, 110 Belgian acute care hospital sites participated in the ECDC and Global-PPS (countrywide participation rate: 81.4%, 28,007 patients). Overall, a crude prevalence of patients with at least one antimicrobial of 27.1% (95% confidence interval (CI) 26.5–27.6%) was found. The most frequently reported indications were pneumonia (23.2%), urinary tract infections (15.2%) and skin and soft tissue infections (11.9%). The reason for antimicrobial use was recorded for 81.9% of the prescriptions, a stop/review date for 40.8% and compliance with local antibiotic guidelines for 76.6%. In the ECDC-PPS, the crude prevalence of patients with at least one HAI was 7.3% (95%CI 6.8–7.7%). Most frequently reported HAIs were pneumonia (21.6%) and urinary tract infections (21.3%). Conclusions HAI and antimicrobial use prevalence remained stable in comparison with the previous PPS (7.1% and 27.4% in 2011 and 2015, respectively). Belgian hospitals should be further stimulated to set local targets to improve antibiotic prescribing and reduce HAI.


Introduction
Healthcare-associated infections (HAIs) and antimicrobial resistance (AMR) are well-known threats in healthcare. Point prevalence studies (PPS) have a wellestablished methodology to measure the prevalence of antimicrobial use and HAIs in hospitals. Results of PPS can be used to evaluate quality indicators, to follow-up antimicrobial stewardship and infection control programs, and to support decision-making [1].
In 2007, a first countrywide PPS of HAIs was conducted in 63 Belgian acute care hospitals. The prevalence of patients with at least one HAI was 6.0%. The most common infections were urinary tract infections (UTIs; 23.9%), lower respiratory tract infections (RTIs; 20.1%) and surgical site infections (SSIs; 14.6%) [2].
In 2011, the European Centre for Disease Prevention and Control (ECDC) organized a first European PPS on HAIs and antimicrobial use in European acute care hospitals. The survey's objectives were to estimate the total burden of HAIs and antimicrobial use in acute care hospitals in the European Union (EU) and therefore provided a standardized tool for hospitals to identify targets for quality improvement [3]. The EU prevalence of patients with at least one HAI was 6.0% (country range 2.3-10.8%). In the participating Belgian acute care hospitals (N = 52 sites), this prevalence of patients with HAIs was 7.1% (95% confidence interval (95% CI): 6.1-8.3%). The EU prevalence of patients receiving at least one antimicrobial agent was 35.0% (country range 21. 4-54.7%). In Belgium, a prevalence of antimicrobial use of 28.9% (95% CI: 26.8-31.1%) was reported [3]. In 2016-2017, a second European PPS was organized by ECDC.
Besides the ECDC-PPS, the Global Point Prevalence Survey of Antimicrobial Consumption and Resistance (Global-PPS) is organized by the University of Antwerp to monitor the ratios of antimicrobial prescribing and resistance in hospitalized inpatients on a worldwide level, with special attention to low-middle income countries. The first Global-PPS was performed in 2015. Worldwide, 34.4% of the included patients received at least one antimicrobial agent [4]. In the included Belgian acute care hospitals (N = 100 hospital sites), a prevalence of 27.4% was found [5]. In 2017, the second Global-PPS was set up.
The aim of this paper is to present the main results of the second ECDC-PPS and Global-PPS of antimicrobial use and HAIs in Belgian acute care hospitals in 2017.

Study design and participation
Two cross-sectional studies, the ECDC and Global-PPS, were simultaneously organized in Belgian acute care hospitals in 2017 by Sciensano and the Belgian Antibiotic Policy Coordination Committee (BAPCOC).
In order to collect data representative for the Belgian hospital population in the ECDC-PPS, a representative sample of hospitals was drawn using a systematic sampling design. For this, a list of all Belgian hospitals was obtained from the Federal Public Service health, Food chain safety and Environment [6]. All acute care hospitals were selected and ranked according to hospital type, total number of beds, region/province and ownership. The total number of hospitals (N = 102 administrative hospital groups (AHG), i.e. hospital sites that are grouped) was divided by the number to be sampled (N = 34) to determine the sampling interval. One substitution per hospital (runner-up) was foreseen in case of refusal of the first selected hospital.
In collaboration with Sciensano, BAPCOC invited the infection control team members and representative(s) of the antibiotic policy group within each hospital to participate in one of the two surveys. The 34 randomly selected hospitals received from Sciensano a personalized invitation and were encouraged to participate in the ECDC-PPS. Because the more complex protocol compared to the Global-PPS and higher workload, financial incentives were given to hospitals conducting the ECDC-PPS (2 euros per included patient). All inclusions in the ECDC-PPS were performed between September and November 2017. The Global-PPS was performed between September and December 2017.

Data collection
The ECDC-PPS was organized in line with ECDC's patientbased protocol for PPS of HAI and antimicrobial use in European acute care hospitals [7]. The protocol of the Global-PPS is available on the Global-PPS website (www.global-pps.com).

ECDC-PPS
Data had to be collected on one single day for each ward in the participating hospitals. The entire data collection could not exceed 2-3 weeks in a single hospital. Accident and emergency wards were excluded. A hospital form collected data on hospital type, size, beds and patients as well as a limited set of structure and process indicators (including the full-time equivalent (FTE) antimicrobial stewardship consultants, interpreted as the time that a consultant/pharmacist is specifically employed and paid for antimicrobial stewardship tasks [7]). For all patients present at the ward before or at 8 a.m. and not discharged from the ward at the time of the survey, a patient form had to be completed. This form collected patient demographic data and risk factors, use of antimicrobial agents and presence of HAIs.
If the antimicrobial (or administration route) changed during the infection episode, the reason for change had to be registered. Possible reasons were escalation (i.e. another antimicrobial was added, the route of administration was switched from oral to parental), de-escalation (i.e. the antimicrobial was switched to a more narrow-spectrum or to a first-line antimicrobial, or other antimicrobials for the same indication were stopped), switch from intravenous to oral administration route, change because of observed or expected adverse effect of the antimicrobial and changes for another or unknown reason [7].
Infections had to be included if they met the definition of an active HAI (associated to an acute care hospital stay). An infection was considered to be active when signs and symptoms of the infection were present on the survey date or when the patient was (still) receiving treatment for that infection on the survey date. The onset of symptoms had to be on day 3 or later (day of admission = day 1) of the current admission or earlier in case the patient presenting an infection was readmitted less than 48 h after a previous admission to an acute care hospital. Exceptions to these inclusion criteria were there for SSIs, Clostridioides difficile infections and infections with invasive devices [7]. Microbiological test results available on the day of the PPS were collected. For a selected group of bug-drug combinations, the antimicrobial susceptibility test results (susceptible, intermediate, resistant or unknown) also had to be reported [7].
Prior to the start of the surveillance period (September 2017), several training days were organized to outline the PPS objectives and methodology to the participating hospitals. Local surveyors had to enter all data into ECDC's HelicsWin.Net software. Thereafter, the local database had to be sent to Sciensano. All individual databases were validated, compiled and transferred to ECDC using their European Surveillance System (TESSy). The data of hospitals that participated in the ECDC-PPS were afterwards converted and imported in the Global-PPS tool (more details on this conversion in Additional file 1).

Global-PPS
Participating hospitals were asked to conduct the survey on one single-day and audit all in-patient wards. All patients present in the ward at 8 a.m. had to be included. Data were collected using two forms, a ward form for the recording of denominators (number of beds and number of admitted patients at 8 a.m. on the day of the PPS) and a patient form for recording detailed antimicrobial prescription (type, dose, administration route, indication, diagnosis) and resistance data for those patients who received at least one antimicrobial on the day of the PPS.
The following ATC groups were included as antimicrobial agents for systemic use: A07AA, D01BA, J01, J02, J04A, J05, P01AB and P01B (antimalarials). Additional antimicrobial quality indicators included 1) the diagnosis being documented in the patient's notes at the start of treatment; 2) the antibiotic prescription being compliant with local guidelines and 3) if a stop or review date of the antimicrobial prescription was documented in the notes. Further, empiric or targeted treatment (based upon microbiology data from a relevant clinical specimen) was recorded. If the treatment choice was determined by available microbiology data, the participant had to indicate if it targeted a multidrug-resistant organism. Data collection forms and definitions on the different variables are available on the Global-PPS website (www.global-pps.com).
The data were entered by the participating hospitals in the freely available web-based application of the Global-PPS. This system allows anonymised data entry, validation and feedback reporting [4]. The complete database is safeguarded at the University of Antwerp. The validated database of June 2018 was used for the analyses in this manuscript.

Data analysis
Where possible, the data of both databases were combined and total results are shown for the Global-and ECDC-PPS (except for data that were collected only in one of the PPS). For the section on HAIs, results of each PPS were separately reported due to the differences in the methodology.
Hospital sites were classified by type (primary, secondary, tertiary or specialized) in accordance with the recommendations of ECDC and based on the list of hospitals of the Federal Public Service Health, Food chain safety and Environment [6,7]. Total results for the Global-and ECDC-PPS were presented per ward specialty (in line with the classification of the Global-PPS); specific results for the ECDC-PPS were presented by patient specialty.
The crude prevalence of patients with at least one HAI or antimicrobial was calculated by dividing the number of patients presenting at least one HAI or antimicrobial by the total number of eligible patients. Patients presenting with multiple HAIs or prescribed multiple antimicrobials on the PPS day were thus counted only once. Prevalences were calculated along with their 95%CI.
Statistical testing was performed, using SAS Enterprise Guide statistical software, version 7.1. Means and standard deviations (SD), ranges and frequencies (%) were calculated where appropriate. The data of the ECDC and Global-PPS on antimicrobial consumption and quality indicators were pooled.

Characteristics of the participating hospitals and included patients
Overall, 110 acute care hospital sites participated in the Global-PPS and ECDC-PPS survey in 2017. Of the random selection (N = 34 AHG), 16 primarily selected hospitals and 6 runner-ups participated in the ECDC-PPS. In total, 28, 007 patients were included, of whom 16,207 patients in the Global-PPS and 11,800 patients in the ECDC-PPS. The characteristics of these hospital sites and their eligible patients are displayed in Tables 1 and 2, respectively.

Antimicrobial consumption
The overarching results of the Global-PPS and ECDC-PPS will be jointly presented. Overall, the crude prevalence of patients with at least one antimicrobial was 27.1% (95%CI: 26.5-27.6%). In total, 7577 patients used one or more antimicrobials on the survey day. These patients had a mean age of 64.8 (SD: ±21.8) years and 51.3% was male. Table 3 presents the crude prevalence by ward specialty, antimicrobial subclass and hospital type. The prevalence by hospital site ranged from 2.2 to 43.3% (interquartile range: 23.0-31.4%).

Antimicrobial quality indicators
Antimicrobial quality indicators are presented by hospital type, ward specialty and indication in Table 4. In general, the reason for antimicrobial use was recorded for 81.9% of the prescriptions and a stop/review date was known for 40.8% of the prescriptions. For antibiotic prescriptions, a compliance with local antibiotic guidelines was reported in 76.6%. For surgical prophylaxis, this compliance was 73.2%. The duration of surgical prophylaxis was in 35.1% of the cases a single dose, in 39.7% 1 day (multiple doses) and in 25.2% more than 1 day. For prolonged surgical prophylaxis (> 1 day), prophylaxis for plastic or orthopedic surgery (30.8%) and prophylaxis for urological surgery (27.7%) were most often registered as diagnosis (only registered in the Global-PPS). In the Global-PPS, 37.9% of therapeutic antimicrobial prescribing (CAI, HAI, LAI) was based on a microbiological result (targeted treatment (N = 1566), whereby following resistant micro-organisms were most often reported: extended-spectrum beta-lactamase (ESBL) producing Enterobacteriaceae (N = 76 patients), third generation cephalosporin resistant Enterobacteriaceae (non-ESBL producing or ESBL status unknown) (N = 39 patients) and methicillin-resistant Staphylococcus aureus (MRSA) (N = 26 patients).
In the ECDC-PPS, a change in the antimicrobial treatment was registered for 24.7% of the antimicrobials (N = 1012): escalation in 425 cases (10.4%), switch from intravenous to oral in 256 cases (6.2%), de-escalation in 237 cases (5.8%), change due to an adverse effect in 21 cases (0.5%), and change for another of unknown reason in 73 cases (1.8%).
The median number of FTEs for antimicrobial stewardship consultants, registered in hospitals participating in the ECDC-PPS (N = 29), was 0.29 per 250 beds (interquartile range 0.20-0.55). These data are only available in the ECDC-PPS whereby patient characteristics were collected for all eligible patients (as opposed to the Global-PPS where this data was only available for patients who received at least one antimicrobial) If data were not collected in the PPS, fields were left blank.

Healthcare-associated infections ECDC-PPS 2017
In the subset of 47 hospital sites (11,800 included patients) that participated in the ECDC-PPS 2017, the crude prevalence of patients with at least one HAI was 7.3% (95%CI 6.8-7.7%, range per hospital site: 0.0-18.1%). The crude prevalences are shown by hospital type and patient specialty in Table 5. For 5.6% (48/856) of the patients with at least one HAI, no antimicrobial medical treatment (prophylaxis excluded) was registered. Based on the indication of antimicrobial prescriptions, the prevalence of patients with a treatment for a HAI was 6.8% (95%CI 6.3-7.2%).
In total, 911 HAIs were registered of which 17.9% (N = 163) were already present on hospital admission. In most cases, the HAI was linked to the current hospital (N = 810, 88.9%) and with the current ward (N = 617, 67.7%). In 20.9% (N = 190) of the cases, the HAI was linked to an invasive device. In Table 6, the main groups of HAIs are displayed by patient specialty. The most common HAIs were pneumonia (N = 197, 21.6%) and UTIs (N = 194, 21.3%).

Discussion
Main results and comparison with previous PPS in Belgian acute care hospitals Prevalence of antimicrobial consumption A crude prevalence of patients with at least one antimicrobial of 27.1% was found. This prevalence is slightly lower than the Belgian results in the previous ECDC-PPS (28.9%) and Global-PPS (27.4%) conducted in 2011 and 2015, respectively. The proportion of medical (5.9%) and surgical prophylaxis (11.2%) decreased in comparison with 2011 (9.0% and 11.8% respectively). The top 3 of most used antimicrobial agents (amoxicillin/beta-lactamase inhibitor, piperacillin/beta-lactamase inhibitor, cefazolin) and the percentage of antimicrobials administered parenterally (64.6%) remained the same as in 2011 and 2015 [3,5]. In the participating European hospitals in the ECDC-PPS 2016-2017, a higher percentage of antimicrobials were administered parenterally (72.8%) [9]. These results also correspond with the results of the national surveillance of antimicrobial consumption in Belgian hospitals (BeH-SAC) [10,11].

Antimicrobial quality indicators
BAPCOC published an action plan 2014-2019 with specific targets for the hospital setting that should be accomplished by 2019 [12]. The first target is that 90% of antibiotic prescriptions should be in line with the local guidelines. In the current study, a compliance of 76.6% was reported. This compliance was lowest in tertiary hospitals (63.5%, but the highest proportion of targeted treatment: 45.7%) and for medical prophylaxis (67.2%). Only in 88.4% of the prescriptions it was indicated that a local guideline was available. In the Global-PPS 2015, guideline availability (90.8%) and compliance (79.5%) was slightly higher [5]. Specifically for surgical prophylaxis, 90% compliance with local guidelines for both the choice and duration of treatment is targeted by BAPCOC [12]. In 2017, the compliance concerning the choice of drug was 73.2%. First-generation cephalosporins (especially cefazolin) were prescribed in 69.2%, followed by penicillins in combination with a beta-lactamase inhibitor (8.8%) and fluoroquinolones (5.9%). The duration of surgical prophylaxis was prolonged (> 1 day) in 25.2%. In 2015, the mean proportion of prolonged surgical prophylaxis cases was 28.2% and cefazolin was prescribed in 62.6% [5]. Penicillins with a beta-lactamase inhibitor and fluoroquinolones are still prescribed in a considerable amount of cases. BAPCOC performed audits on surgical prophylaxis in hospitals in 2013 and 2017; choice and duration compliance are expected to further improve. More detailed analyses are needed to allow target setting for specific indications.
The action plan of BAPCOC stated that the indication of the prescription should be available in the medical file in 90% [12]. This study shows that this target is not yet achieved (reason for antimicrobial use was recorded in 81.9%). Finally, post-prescription review is also an important antimicrobial quality indicator to prevent unnecessary prolonged antimicrobial use [13]. The Global-PPS showed the availability of a stop/review date in only 40.8% of the prescriptions.

Prevalence of healthcare-associated infections
Based on the ECDC-PPS 2017 data, a crude prevalence of patients with at least one HAI of 7.3% was detected in Belgian acute care hospitals. The highest prevalences were found in tertiary hospitals (9.1%) and on ICU (20.9%). In comparison with the ECDC-PPS in 2011, this prevalence remained stable (7.1%) [3]. The most frequently reported HAIs were pneumonia, UTIs and SSIs, which is in line with the previous PPS [3].

Comparison with other European countries
The overall prevalence of antimicrobial consumption in Belgian acute care hospitals (27.1%) is lower than reported for the participating European hospitals in the ECDC-PPS 2016). In addition, the amount of prolonged surgical prophylaxis (> 1 day) in Belgian acute care hospitals (25.2%) was clearly lower than in the participating European hospitals (54.2%, country range 19.8-95.0%) [9]. Belgian acute care hospitals reported a  [9]. In contrary, the prevalence of patients with at least one HAI in Belgian acute care hospitals (7.3%) remains clearly higher than in other European hospitals. A weighted prevalence of HAI of 5.5% (95%CI 4.5-6.6%) was reported in the participating European hospitals in the ECDC-PPS 2016-2017 [14]. In Scottish hospitals which performed the ECDC-PPS in 2016, a prevalence of antimicrobial use of 35.7% (95%CI 34.2-37.2%) and HAI prevalence of 4.6% (95%CI 4.1-5.1%) were detected in acute adult patients [15].

Strengths and limitations
In 2017, two PPS were performed in Belgian acute care hospitals in line with the international standardized methodology of ECDC-PPS and Global-PPS. The added value of combining the data of the two surveys is that it resulted in a large database with a participation rate of more than 80% of all Belgian acute care hospitals. Moreover, each PPS system has its own added value (e.g. ECDC-PPS: more detailed data on HAIs, Global-PPS: other quality indicators), both at local and national level, resulting in a more complete picture on the current practice in Belgian acute care hospitals and allowing international comparisons. To our knowledge, it is the first time that data of both PPS using a similar methodology are combined. All participating hospitals already received an individual feedback of their prevalence data, benchmarked with national results. Participation, if organised countrywide, is also recorded as a mandatory quality indicator to improve infection control in Belgian hospitals [16]. Some limitations should be acknowledged. As this was a cross-sectional survey, only prevalences can be reported and patients were not followed-up in time. Prevalence study only show a snapshot of the situation, therefore they are not ideal for measuring for example AMR. However, in addition to the PPS, different (mandatory and voluntary) surveillances on AMR and HAIs are conducted in Belgian hospitals (e.g. surveillances on blood-stream infections, Clostridioides difficile, SSI, MRSA, vancomycin resistant enterococci (VRE), multi-resistant Gram negative bacteria) [10]. Secondly, the results were not corrected for patient case mix or institutional factors. Therefore, comparison with the results of different PPS should be interpreted carefully as different hospitals participated each time. Detected differences between several PPS might be explained by differences in case mix. In addition, the definition of the type of hospitals might have been different in past PPS. Finally, although we managed to merge the data of both PPS in the best way (see Additional file 1), some differences in the protocols of the ECDC-PPS and Global-PPS could have had a small influence on the results (e.g. antivirals (J05, 95 prescriptions: 1.03%) and antimalarials (P01B, 3 prescriptions: 0.03%) were only included in the Global-PPS). As only one hospital participated in both surveys, a direct comparison of the results obtained from both study methods was not assessed here.

Future perspectives
It is recommended that PPS are repeatedly performed to follow-up the evolution over time and monitor the impact of antimicrobial stewardship and infection prevention programmes [1]. The Global-PPS tool is continuously available and hospitals themselves can decide how often they conduct a PPS. ECDC aims to organize a PPS every 5 years (third ECDC-PPS probably planned in 2022). On a national level, a fixed timeinterval (e.g. every 2 years) should be defined to be able to provide data for benchmarking and to evaluate targets for example as set up in the 2014-2019 action plan of BAPCOC. Since 2018, Belgian hospitals are partly financed based on the level of quality in their hospital (Pay for Performance project), and participation in a PPS is one of the conditions for the financing [17]. There is still a large range in the prevalence of antimicrobial consumption (2.2 to 67.3%) and of HAIs (0.0 to 18.1%) between hospital sites. Hospitals with outlying results should further be targeted to help the local antibiotic policy teams develop specific antimicrobial stewardship and infection prevention programmes to improve.
In general, concerning the antimicrobial consumption, the high prescribing of 'Fluoroquinolones' (J01MA) is a concern and should be a target for intervention. In addition, the results of the discussed antimicrobial quality indicators should be further improved to reach the BAPCOC targets (90% for all indicators). Moreover, the number of patients who develop a HAI in Belgian acute care hospitals remains high. The reasons for these high rates of HAIs should be further investigated.

Conclusions
In comparison with previous PPS, the prevalence of antimicrobial use and HAI and most results for the antimicrobial quality indicators remained status quo. Belgian hospitals should be further stimulated to regularly participate in a PPS and to set local targets for improving antibiotic prescribing and reducing HAI.
Additional file 1: Table S1. Overview how the ECDC-PPS data were converted in the Global-PPS database (Belgium, acute care hospitals, 2017).