In this century, antimicrobial resistance (AMR) has become a serious threat and a major concern to global public health.1,2 This is particularly relevant in hospital settings where infections due to multidrug-resistant (MDR) microorganisms are more prevalent, and are associated with higher morbidity, mortality, and financial costs.3 AMR-associated infections are considered costly yet avoidable; the United States (US) Centers for Disease Control (CDC) estimates that the direct healthcare costs associated with AMR surpass $20 billion per year in US, with an additional cost to society, for lost productivity, of $35 billion.4 Middle-income countries (Brazil, Russia, India, China, and South Africa; also referred to as BRICS) accounted for three-quarters of the global increase in antibiotic consumption in the past decade.5 However, only a small number of studies have examined strategies to improve antibiotic consumption in these countries.6,7

The high prevalence of extended-spectrum beta-lactamase (ESBL)-producing bacteria in hospitals has been observed worldwide. The prevalence of ESBL-producing Escherichia coli and Klebsiella pneumoniae reached 17.6% and 38.9% in Europe, 8.5% and 8.8% in US, 40.8% and 21.5% in the Asia-Pacific region, and 26.8% and 37.7% in South America, respectively.8–10 Consequently, consumption of “last-resort” antibiotics, such as carbapenems and polymyxins, has increased. Between 2000 and 2010, the global consumption of carbapenems and polymyxins increased by 45% and 13%, respectively,5 followed by an increasing trend until 2015.11 However, antibiotic consumption is a determinant driver for AMR.12–16

Strategies such as antimicrobial stewardship programs (ASP) appear to be an efficient way to improve clinical and microbial outcomes; such programs aim to reduce the inadequate use of antibiotics and the associated expenditure.17,18 The aim of this study was to evaluate antibiotic use, the associated costs, and the bacterial susceptibility profiles after implementation of a broad-spectrum beta-lactam-sparing ASP.

This study evaluated the impact of an ASP focused on reducing the use of broad-spectrum beta-lactams in a hospital in a middle-income country. The program was associated with an 11.33% reduction in the overall antibiotic use. DOT with broad-spectrum beta-lactams decreased by an average of 24.58%, with the change observed immediately after implementation of the ASP. In contrast, use of beta-lactam-sparing antibiotics gradually increased during the intervention period, resulting in a total increase of 45.54% by the end of the 12-month intervention. Costs attributed to antibiotic use were reduced by $3934.41 per month in the intervention period (an estimated one-year saving of $47,212.92). A downward trend in monthly antibiotic expenditure was noted prior to the formal initiation of ASP, attributed to the withdrawal of intravenous ciprofloxacin from antibiotics formulary three months before the ASP was implemented (supplementary material 2). However, a more substantial drop was observed immediately after ASP implementation, which was sustained throughout the intervention period. There was an increase in fluoroquinolone use, although expenditure on fluoroquinolones decreased in the intervention period owing to the increased consumption of oral formulations. A higher proportion of drug-susceptible Klebsiella spp. and aminoglycoside-susceptible A. baumannii were isolated during the intervention period.

Prior to this study our hospital had limited experience with formal programs promoting the rational use of medicines, beyond standard antibiotic prescription protocols without audit or management. We speculated that this was the reason why baseline broad-spectrum antibiotic use was so high and believe that this provided an excellent opportunity to moderate usage. Our baseline mean monthly consumption of carbapenems was 51.28 DOT/1000 patient-days, almost double than that reported in hospitals in the US, and our use of third and fourth-generation cephalosporins was more than double.22 As observed in similar studies in high-income countries, we have demonstrated success in reducing the use of selected antibiotics and their associated costs. Our intervention resulted in an 11.3% reduction in antibiotic use and a 53% ($3934.41) reduction in associated monthly expenditure, which was sustained throughout the study period. Similarly, Campbell et al. reduced antibiotic use in a hospital surgical ward by 12% by applying a non-restrictive policy; this was accompanied by a 35–41% reduction in antibiotic costs.23 In a study by Borde et al., after an intensive education program aimed at discouraging the use of quinolones and cephalosporins, the associated observed reduction was 14.2% in the medical wards, with a saving of €135,000 annually, excluding personnel costs.24 A study by Jenkins et al., using a semi-restrictive methodology for broad-spectrum antibiotics similar to ours, demonstrated a 14.2% reduction in total antibiotic use and a 30% reduction in the use of broad-spectrum agents; this impact was observed shortly after the start of the program and was associated with a 27.9% decrease in antibiotic expenditure.25

In a systematic review and meta-analysis, Karanika et al. evaluated the economic outcomes of 26 studies and revealed a 19.1% reduction in total antibiotic consumption and 26.6% reduction in the use of restricted antibiotics,26 similar to our reduction of 24.6%. Although there is an absolute need to reduce antibiotic consumption in BRICS countries, so far only a small number of studies have been published on this topic from low to middle-income countries. For example, in a study conducted in a cardiology hospital in Brazil, Magedanz et al. demonstrated a 25% and 69% reduction in antibiotic consumption and cost, respectively; this study attributed the success to collaboration between the clinical pharmacist and infectious diseases specialist and to the use of a non-restrictive policy.27 We observed that the use of restricted antibiotics in our study fell immediately after implementation of the ASP and that this reduction was maintained throughout the intervention period. This change was consistent with those generally observed in studies that adopt semi-restrictive policies,28,29 where the restriction results in an abrupt reduction in consumption, and auditing and feedback allow this level of use to be sustained over time.

The success of ASPs appears to be primarily related to the baseline level of consumption, in other words, the opportunity for reduction. Furthermore, the antibiotics targeted are an additional factor associated in the success of ASPs. Even small reductions in the consumption of last-resort and broad-spectrum antibiotics have an impact on the financial outcome.

There were no significant changes in the proportion of susceptibility of E. coli, Enterobacter spp., and Pseudomonas aeruginosa. We also expected an improvement in susceptibility to beta-lactams in the majority of bacteria strains, especially for carbapenems, for which reduction between periods was the greatest. Despite a dramatic increase in the consumption of fluoroquinolones, we did not see any worsening in the resistance profile of E. coli and K. pneumoniae, which has been reported in some studies.30,31 It is difficult to relate these results only to the control of antimicrobials, and there is huge variation between study models, consumption of different drugs, and follow-up time. Therefore, it is unclear how the ASP implementation impacts bacterial susceptibility profiles.13,16,25,32 Another potential source of bias related to the change is the method for laboratory identification of bacteria, which was previously performed manually by means of culture and disk-diffusion antibiograms, whereas in the intervention period, by automated methods. The susceptibility profile of Acinetobacter spp. to aminoglycosides increased from 31% to 73% with implementation of the ASP. This finding has not been reported elsewhere in the literature, but we proposed the following hypothesis. The carbapenem-resistant Acinetobacter spp. in our hospital is endemic and associated with OXA-23 carbapenemase.33 Resistance to aminoglycosides can be mediated by methylases, other aminoglycoside-modifying enzymes, and efflux pumps. The main pump associates with amikacin resistance is the adeABC.34 The adeABC gene is overexpressed by carbapenem.35 Thus, with a decreased use of carbapenem, it is possible that adeABC gene is not overexpressed and susceptibility to amikacin is therefore increased.

Although the safety of aminoglycosides has been discussed widely in the literature from the perspective of nephrotoxicity, the exchange of beta-lactams for this class was initially encouraged by the favorable susceptibility profiles of Gram-negative bacteria to amikacin and gentamicin. Furthermore, previous studies have corroborated the safety of these drugs when used in adjusted doses and with serum monitoring.36,37 In addition, a new meta-analysis highlighted the association of acute renal injury with patients receiving concomitant piperacillin-tazobactam and vancomycin, which reinforced our exchange strategy.38

This study has some limitations. First, the collection of retrospective data limits the discussion of results, especially in the case of actions performed in the pre-intervention period. Second, the ecological study design allowed us to hypothesize on ASP outcomes; however, we cannot say that the actions of the program were the only driver of results. This is especially relevant regarding bacterial susceptibility profiles, which are already known to be affected by numerous factors other than the level of antibiotic use.39 Nonetheless, we used the ITS model, which is considered a powerful tool in observational studies addressing longitudinal analyses of systemic interventions, primarily in the field of medication consumption and costs.20 The most recent review in the Cochrane Library concerning outcomes from the optimization of antibiotic prescribing included 138 studies with an ITS design, highlighting the importance of this model in the production and compilation of similar reports.

In summary, a semi-restrictive ASP based on sparing extended-spectrum beta-lactams was an effective strategy for reducing targeted antibiotic consumption in a hospital setting and maintaining susceptibility profiles in Gram-negative bacteria. This study also fits with the global efforts to reduce inappropriate prescription of last-resort antibiotics, mainly in middle-income countries, and demonstrated a cost-saving strategy that was achievable with a small, specialized team.

This research did not receive any specific grants from funding agencies in the public, commercial, or not-for-profit sectors.

The authors declare no conflicts of interest.

Tiago Zequinao, Pharm – wrote manuscripts and statistical analysis.

Juliano Gasparetto – revised the manuscript and database evaluation.

Dayana dos Santos Oliveira – database evaluation and revised the manuscript.

Gabriel Takahara Silva – data evaluation.

João Paulo Telles - database evaluation and revised the manuscript.

Felipe Francisco Tuon – wrote and revised the manuscript.