Antibiotic consumption and healthcare-associated infections in a tertiary hospital in Belgrade, Serbia from 2011 to 2016

Introduction: Healthcare-associated infections (HAIs) and irrational use of antibiotics in healthcare settings are major global public health concerns. Surveillance of HAIs in intensive care units (ICU), surgical-site infections (SSIs), and Clostridium difficile infections (CDIs), together with implementation of antibiotic stewardship, are cornerstones of hospital infection prevention programs. The aim of this study was to evaluate antibiotic consumption, especially of broad spectrum antibiotics, in relation to HAI incidence density (ID). Methodology: The study was conducted from 2011 to 2016 in a tertiary hospital, the Military Medical Academy (MMA), in Belgrade, Serbia. Through regular hospital surveillance we identified all patients with a new HAI. Data on consumption of antibacterials for systemic use were expressed as defined daily dose per 100 bed days (DDD/100 BD). Results: The highest incidence density (ID) of HAI was observed among patients in surgical ICUs (47.2 per 1000 patient-days), while the highest incidence rate among SSI was 3.7%. Moreover, the highest ID of CDI in medical patients was 6.2 per 10,000 patient-days, while in surgical patients it was 4.3 per 10,000 patient-days. The most frequently used antibiotics were cephalosporins, aminoglycosides and carbapenems (16.0 ± 2.3, 4.8 ± 0.7, 4.3 ± 0.7 DDD/100 BD, respectively). There was no significant correlation between consumption of any groups of antibiotics and ID of CDI in medical and surgical patients. Conclusion: The multidisciplinary healthcare team would have crucial importance in the implementation of the antibiotic stewardship program in order to decrease unnecessary exposures of patients treated in healthcare settings.


Introduction
Healthcare-associated infections (HAIs) and irrational use of antibiotic in healthcare settings are major global public health concerns [1]. The intensive care unit (ICU) is often the epicenter of emerging problems of HAIs and antimicrobial resistance in a hospital. Patients admitted to ICU are at higher risk of HAI due to both intrinsic (age, underlying diseases, immunosuppression) and extrinsic (use of invasive devices such as mechanical ventilation, central vascular catheter,) risk factors (RFs) [2]. A systematic review and meta-analysis of the burden of endemic HAI in developing countries showed that pooled overall HAI infection density in adult ICU was at least three times higher than those reported from the US. Strikingly higher surgical-site infection (SSIs) rates, the most frequent HAI in surgical patients, were also observed in developing countries [3]. In contrast, a systematic review describing the epidemiology and management of Clostridium difficile infection (CDI) in developing countries suggested that the rate of community associated and healthcare-associated (HA) CDI appeared to be lower in developing countries than in developed countries, yet RFs appeared to be broadly similar between these two populations [4]. Antibiotic exposure represents a principal RF for CDI, in both developed and developing countries [5,6].
Antibiotics are among the most frequently used drugs in hospitalized patients. Each antibiotic has a different unit dose of daily administration; a specific standardized method should be used in the evaluation of in-hospital use. The World Health Organization (WHO) recommends the use of the Anatomical Therapeutic Chemical/Defined Daily Dose (ATC/DDD) index as a method to express antibiotic consumption [7].
Appropriate antibiotic use in hospitals should ensure effective treatment but reduce unnecessary prescriptions [8]. Severe infections with multidrugresistant pathogens present a medical challenge and a financial burden for hospitals. Critically ill patients are at very high risk of developing severe HAI in ICU, with incidence density (ID) 5-10-fold higher than in general medical wards [9,10]. The use of broad spectrum antibiotics is a particular problem. Characteristics such as the type of healthcare, knowledge of local common pathogens and resistance patterns as well as patient RFs for colonization or infection with multidrug-resistance pathogens should be used to select empirical broad spectrum antibiotics [11].
Surveillance of HAIs in ICU, of SSIs and CDIs together with implementation of antibiotic stewardship are cornerstones of infection prevention programs in modern hospitals.
The aim of this study was to evaluate the antibiotic consumption, especially the broad spectrum ones, and investigate its relation to HAI ID in a tertiary hospital in Serbia.
The surveillance of SSIs was conducted among general and orthopedic surgery patients, followed by surveillance of vascular and cardiac surgery, thoracic surgery, plastic surgery and burns, urology and maxillofacial surgery patients. The wound class was assigned after operative procedure completion based on actual occurrences as clean, clean/contaminated, contaminated, and dirty/infected [13]. For diagnosis of SSI, the CDC criteria were used [14].
Patients from all departments were included for the surveillance of CDI. A CDI case was defined as any hospitalized patient with laboratory confirmation of a positive toxin assay or Clostridium difficile nucleic acid amplification assay results together with diarrhea (≥ 3 daily in a 24-hours' period with no other recognized cause) or visualization of pseudomembranes on sigmoidoscopy, colonoscopy, or histopathologic analysis [15,16]. Microbiological testing was performed at the Institute of Medical Microbiology at the MMA.
Only the first episode of HAIs, SSIs and CDIs was included for patients who had more than one infection during the study period.
The ID of HAIs in ICU was calculated as the number of HAI per 1000 patient-days. The IR of SSI was defined as the number of SSIs per 100 operative procedures. The ID of CDI was defined as the number of HAI CDI caused per 10,000 patient-days in medical and surgical patients.
Relative numbers were expressed as percentages. In the case of continuous data, variables were presented as a mean value ± standard deviation (SD). Pearson correlation analysis was used to establish the relationship between different antibiotic groups and ID of CDI in medical and surgical patients. Statistical significance was set at a level of p<0.05. Complete statistical analysis of data was done using SPSS Statistics 18 (SPSS Inc., Chicago, IL, USA).

Healthcare-Associated Infections
The highest ID of HAI was observed in patients in surgical ICUs with a peak in 2015 (47.2 per 1000 patient-days). There was a trend of increasing ID of HAIs in the medical and toxicological ICU during the study period. The neurological ICU was founded in 2014; a peak of HAIs was registered in 2015 (23.4 per 1000 patient-days) ( Figure 1).
From January 1st, 2011 to December 31st, 2016, a total of 35,995 different operative procedures were evaluated, with an overall cumulative IR of 2.9% (1.048) SSIs.
The overall ID of CDIs was highest in 2016 (4.7 per 10,000 patient-days). The highest ID of CDI among medical patients was observed in 2016 (6.2 per 10,000 patient-days), while in surgical patients it occurred in 2013 (4.3 per 10,000 patient-days) ( Figure 3).

Antibiotic consumption
The highest consumption was detected in the years 2013 and 2015 (46.5 and 46.0 DDD/100 BD, respectively) and the lowest consumption in the year 2016 (33.4 DDD/100 BD). In 2016 there was a decrease in consumption of cephalosporins of 25% compared with 2015. Overall hospital consumption of antibiotics over a six-year period is shown in Table 1.
The most frequently used antibiotics from 2011 to 2016 were cephalosporins, followed by aminoglycosides and carbapenems (16.0 ± 2.3 DDD/100 BD, 4.8 ± 0.7 DDD/100 BD and 4.3 ± 0.7 DDD/100 BD, respectively). The highest consumption of cephalosporins was in 2011 (18.6 DDD/100 BD), with a rate of 43.5% of total consumption of antibiotics in that year, followed by a steady decrease in the following years. The lowest consumption was in the year 2016 (12.4 DDD/100 BD), at 37.1% of total consumption of antibiotics in that year. The trend in consumption of aminoglycosides was the same as for overall antibiotics consumption. It was highest in 2013 (5.6 DDD/100 BD), at 12.1% of total consumption of antibiotics and lowest in 2016 (4.1 DDD/100 BD) at 12.4%. Carbapenem consumption was different in comparison to the previous two groups. The highest consumption was recorded in 2015 (5.0 DDD/100 BD), at 11.0% of antibiotic consumption, while the lowest consumption was in 2011 (3.1 DDD/100 BD), at 7.3% ( Figure 4). Meropenem was the mostly commonly used carbapenem (average consumption 2.1 ± 0.6 DDD/100 BD).
Antibiotics were grouped according to their use for patients with different levels of risk of developing CDI: high-risk antibiotics (cephalosporins, carbapenems, fluoroquinolones and clindamycin), medium-risk antibiotics (penicillins, aminoglycosides, macrolides and sulfametoxazol/trimethoprim) and low-risk antibiotics (tetracycline). The consumption of the three groups of antibiotics in DDDs per 100 BD in a six-year period is shown in Table 2.
The consumption of the 3rd generation cephalosporins (ceftriaxone and ceftazidime) was 66.8% of all cephalosporins used in a six-year period. Overall consumption of cephalosporins in defined daily doses (DDDs) according to Anatomical Therapeutical  Chemical classes per 100 bed days (BD) is shown in Figure 5.
On the other hand, several broad spectrum antibiotics had a total consumption less than 5 DDD/100 BD during the six-year period. The consumption of glycopeptides (vancomycin and teicoplanin) was in the range of 1.3 DDD/100 BD in 2016 to 3.0 DDD/100 BD in 2015, while teicoplanin use was less than 1.00 DDD/100 BD.
Colistin consumption was at 0.1 DDD/100 BD from 2011 to 2014, but there was an increasing trend in the two following years and a total consumption of 0.2 DDD/100 BD.
Linezolid use was recorded only during three years (in 2014 and 2016 it was 0.1 DDD/100 BD and in 2015 twice that).
Correlations between consumption of antibiotics classified according to the risk for patients to develop Clostridium difficile infection (CDI) and incidence density (ID) of CDI in medical and surgical patients are shown in Table 3. There was no statistically significant correlation between any groups of antibiotics. However, the 3rd generation cephalosporins showed correlation with the highest value of total consumption increase ID CDI in surgical patients (Pearson correlation: 0.256).

Discussion
During the six-year period from 2011 to 2016, surveillance on HAIs was conducted in patients at the MMA hospital in Belgrade; antibiotic consumption was also measured. Like data obtained from the study conducted in another tertiary hospital in Serbia [18], our investigation showed that cephalosporins were the most frequently prescribed antibiotics, followed by aminoglycosides and carbapenems.
Although we noticed a decrease in cephalosporin use in 2016, the use of the 3rd generation cephalosporins was constant over the study period. Our results are in accordance with studies from other countries, where cephalosporins were the most frequently used antibiotics in hospitals, ranging from 30% in Chinese tertiary hospitals to 57% in Turkish hospitals [19][20][21]. Reports of antibiotic consumption in  European hospitals during 2011-2015 showed that consumption of cephalosporins and other beta-lactams included carbapenems ranged from 7% in the United Kingdom to 54% in Bulgaria [22]. National data on antibacterial consumption in Australia for the five-year period (2012-2016) revealed the greater usage of broad spectrum cephalosporins (ceftriaxone, ceftazidime, cefepime and cefotaxime) in larger hospitals might be explained by a more complex casemix [23]. During 2011, Ju et al. reported that rate of SSI was the highest among cases that were classified as dirty/infected (8.5%) and the lowest in the cases of the wounds that were clean (1.8%) [24], which is similar to our observations. Third generation cephalosporins, especially ceftriaxone, have generally not been recommended for surgical antimicrobial prophylaxis [25][26][27], but they have been widely accepted as the most common drugs in surgical antimicrobial prophylaxis in our hospital. Administration of surgical antimicrobial prophylaxis less than 30 minutes before incision has been routine practice at the MMA. The recommended duration of surgical antimicrobial prophylaxis is less than 24 hours for most procedures (except cardiothoracic procedures). However, in every day practice during clean/contaminated, contaminated and dirty/infected operative procedures, MMA surgeons extended surgical antimicrobial prophylaxis to more than 24 hours. The first EU-wide, ECDC-coordinated point prevalence survey of HAIs and antimicrobial use in acute care hospitals showed that surgical antimicrobial prophylaxis was prolonged for more than one day in 59.2% of cases [28]. However, the last CDC Guideline for the Prevention of SSI, published in 2017, pointed out that for clean and clean-contaminated procedures, additional prophylactic antimicrobial agent doses should not be administered after the surgical incision is closed in the operating room, even in the presence of a drain [29].
Decreased use of fluoroquinolones was noticed in 2016. The most frequently used was ciprofloxacin, while moxifloxacin has not been used yet due to the limited number of standard indications. A report from Australia showed that fluoroquinolone use had a downward trend over the period 2012-2016 [23], while in MMA the decrease of 37% was recorded in 2016 in comparison to 2015. Data from nine regional tertiary hospitals in Ireland for 2011-2012 showed that fluoroquinolone use was 5.8 DDD/100 BD in 2011 and 5.9 DDD/100 BD in the 2012 [30]. In the same period, the consumption of the same group in MMA was lower  [31]. Recently, authors from England reported that national fluoroquinolone and cephalosporin prescribing practice correlated highly with the incidence of CDI, in contrast to total antibiotic prescribing. They also concluded that restriction of fluoroquinolone prescribing should be a cornerstone in the control of epidemic CDI in the UK and worldwide [32].
We did not find a correlation between fluoroquinolone consumption (nor any other group of antibiotics) and ID of CDIs. The only association established between ID of CDI and an individual group of antibiotics was with the 3rd generation cephalosporin in surgical patients. The overall ID of CDI was highest in 2016 at 4.7 per 10,000 patient-days in our hospital, which is lower than the mean ID of 7.0 of HA CDI (country range 0.7-28.7) in a European, multicentre, prospective, biannual point prevalence study of CDI in hospitalized patients with diarrhea (EUCLID) [33]. We also observed the differences in ID of CDI between medical and surgical patients. Silva-Velazco et al. reported that although surgical patients tended to have more difficult clinical presentation of CDI, overall they did better than medical patients. However, future studies focusing on modifiable RFs for each group are needed [34].
In order to improve the quality of especially broad spectrum antibiotic use, a multidisciplinary healthcare team (infectious disease specialist, epidemiologist, clinical pharmacologist, hospital pharmacist, clinical microbiologist) was formed in our hospital in 2008. The main goal of this team is to monitor the use of antibiotics and to make that more appropriate. The team determined the list of restricted antimicrobial agents as well as the criteria for their use combined with an approval system. In comparison to analysis of antibiotic consumption in the previous ten-year period (2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010) in MMA, when the average consumption was 56.6 ± 3.01 DDD/100 BD, we have noticed a 25% decrease in consumption of antibiotics, due to the efforts of the multidisciplinary team to reduce irrational use of antibiotics in our hospitals, especially the restricted ones [35].
The highest consumption of carbapenems was recorded in 2015, with a total rate of 11.0%. It is not surprising that in the same year the highest ID of HAI was observed in patients in surgical ICU and neurological ICU. The most frequently used was meropenem, which has become a key restricted-line antibacterial because of its superior activity against Pseudomonas species and lower incidence of neurotoxicity.
The median consumption rate of tygecycline in our hospital was similar to that reported from Ireland (3.3 DDD/100 BD in Ireland vs. 2.9 DDD/100 BD in MMA in the year 2011; 3.5 DDD/100 BD in Ireland vs. 3.7 DDD/100 BD in MMA in the year 2012) [30]. Its current position in the context of rational antibiotic prescribing policy in our hospital is not yet defined.
In MMA, colistin is an important antibacterial in the treatment of HAIs caused by carbapenemase-producing multidrug-resistant gram-negative microorganisms, where meropenem is ineffective. Since it was registered for the Serbian market in 2013, it was not often used up to 2014, with a slight increase in the following two years. This result reflects the fears of MMA doctors about its serious adverse effects.
Linezolid is reserved for complex infections caused by multidrug resistant gram-positive organism including vancomycin-resistant Enteroccoci (VRE).
Parenteral colistin, teicoplanin and linezolid usage remained low, but showed a possible trend upwards that requires ongoing monitoring. The very low usage teicoplanin may be due to its higher cost in comparison to vancomycin in Serbia.

Strengths and limitations
Our study has several limitations. First, it was a single-center study. The consumption of antibiotics may not be representative for other hospitals in the country. Second, we did not calculate the resistance ratio of microorganisms which were the cause of HAI during the study period. One of the limitations was the DDD methodology for calculating use of antibiotics in our research. The use of prescribed daily doses (PDDs) is nowadays widely used for measuring antibiotic consumption in hospitals because it will give the amount of a drug that has actually been prescribed. Some authors think that there are discrepancies between the two methods [17,36]. Obviously, additional research is needed to determine which method should be used for calculation of antibiotic usage. Finally, we have not taken into account the oral formulations of antimicrobial drugs, which will be considered in future investigations.
The strength of this study is the large material covering the consumption of all antibiotics used in the MMA and the emergence of HAI in ICU, SSI and CDI during six years. This long surveillance period gives good insight into the rate of HAIs and the consumption of antibiotics in a tertiary care hospital in a small, poor, developing country with the problems resulting from economic transition in the healthcare system.

Conclusions
Our analysis showed that cephalosporins were the most frequently prescribed antibiotics, followed by aminoglycosides and carbapenems in a tertiary hospital in Belgrade, Serbia. Surgical antimicrobial prophylaxis is one reason for the high consumption of cephalosporins in MMA. The multidisciplinary healthcare team had very important role in monitoring of the use of antibiotics and would have crucial importance in the implementation of the antibiotic stewardship program in order to decrease unnecessary exposures of patients treated in MMA.