Antimicrobial Stewardship Programs: Effects on Clinical and Economic Outcomes and Future Directions
Results
Antimicrobial Usage
The most widely studied aspect of ASPs in the current review was the effect of ASP interventions on antimicrobial consumption and use. Three systematic reviews [22–24] showed improved antibiotic prescribing practices and reduced consumption rates overall, as did several studies inside and outside the intensive care unit (ICU) [25–31].One study found an insignificant declining usage trend [32]. An important underlying facet of this observation is that even as total antibiotic consumption decreases, certain antibiotic and antibiotic class consumption may increase. This is evident in several studies, which showed that as aminoglycoside, carbapenem, and β-lactam-β-lactamase inhibitor use increased, clindamycin (1 case), glycopeptide, fluoroquinolone, and macrolide use decreased [27,28,30]. A potential confounding factor relating to decreased glycopeptide use in Bevilacqua et al [30] was that there was an epidemic of glycopeptide-resistant enterococci during the study period, potentially causing prescribers to naturally avoid it. In any case, since the aim of ASPs is to encourage a more judicious usage of antimicrobials, the observed decreases in consumption of those restricted medications is intuitive. These observations about antimicrobial consumption related to ASPs are relevant because they putatively drive improvements in clinical outcomes, especially those related to reduced adverse events associated with these agents, such as the risk of C. difficile infection with certain drugs (eg, fluoroquinolones, clindamycin, and broad-spectrum antibiotics) and prolonged antibiotic usage [33–35]. There is evidence that these benefits are not limited to antibiotics but extend to antifungal agents and possibly antivirals [22,27,36].
Utilization, Mortality, and Infection Rates
ASPs typically intend to improve patient-focused clinical parameters such as hospital LOS, hospital readmissions, mortality, and incidence of infections acquired secondary to antibiotic usage during a hospital stay, especially C. difficile infection. Most of the reviewed evidence indicates that there has been no significant LOS benefit due to stewardship interventions [24–26,32,37], and one meta-analysis noted that when overall hospital LOS was significantly reduced, ICU-specific LOS was not [22]. Generally, there was also not a significant change in hospital readmission rates [24,26,32]. However, 2 retrospective observational studies found mixed results for both LOS and readmission rates relative to ASP interventions; while both noted a significantly reduced LOS, one study [38] showed an all-cause readmission benefit in a fairly healthy patient population (but no benefit for readmissions due to the specific infections of interest), and the another [29] showed a benefit for readmissions due to infections but an increased rate of readmissions in the intervention group overall. In this latter study, hospitalizations within the previous 3 months were significantly higher at baseline for the intervention group (55% vs. 46%, P = 0.042), suggesting sicker patients and possibly providing an explanation for this unique observation. Even so, a meta-analysis of 5 studies found a significantly elevated risk of readmission associated with ASP interventions (RR 1.26, 95% CI 1.02–1.57; P = 0.03); the authors noted that non–infection-related readmissions accounted for 61% of readmissions, but this was not significantly different between intervention and non-intervention arms [37].
With regard to mortality, most studies found no significant reductions related to stewardship interventions [22,24,26,29,32]. In a prospective randomized controlled trial, all reported deaths (7/160, 4.4%) were in the ASP intervention arm, but these were attributed to the severities of infection or an underlying, chronic disease [25]. One meta-analysis, however, found that there were significant mortality reductions related to stewardship guidelines for empirical antibiotic treatment (OR 0.65, 95% CI 0.54–0.80, P < 0.001; I2 = 65%) and to de-escalation of therapy based on culture results (RR 0.44, 95% CI 0.30–0.66, P < 0.001; I2 = 59%), based on 40 and 25 studies, respectively [39]; but both results exhibited substantial heterogeneity (defined as I2 = 50%–90% [40]) among the relevant studies. Another meta-analysis found that there was no significant change in mortality related to stewardship interventions intending to improve antibiotic appropriateness (RR 0.92, 95% CI 0.69–1.2, P = 0.56; I2 = 72%) or intending to reduce excessive prescribing (RR 0.92, 95% CI 0.81–1.06, P = 0.25; I2 = 0%), but that there was a significant mortality benefit associated with interventions aimed at increasing guideline compliance for pneumonia diagnoses (RR 0.89, 95% CI 0.82–0.97, P = 0.005; I2 = 0%) [37]. In the case of Schuts et al [39], search criteria specifically sought studies that assessed clinical outcomes (eg, mortality), whereas the search of Davey et al [37] focused on studies whose aim was to improve antibiotic prescribing, with a main comparison being between restrictive and persuasive interventions; while the difference may seem subtle, the body of data compiled from these searches may characterize the ASP effect of mortality differently. No significant evidence was found to suggest that reduced antimicrobial consumption increases mortality.
Improving the use of antimicrobial agents should limit collateral damage associated with their use (eg, damage to normal flora and increased resistance), and ideally infections should be better managed. As previously mentioned, one of the concerns with antibiotic usage (particularly fluoroquinolones, macrolides, and broad-spectrum agents) is that collateral damage could lead to increased rates of C. difficile infection. One meta-analysis showed no significant reduction in the rate of C. difficile infection (as well as overall infection rate) relative to ASPs [22]; however, this finding was based on only 3 of the 26 studies analyzed, and only 1 of those 3 studies utilized restrictions for flouroquinolones and cephalosporins. An interrupted time series (ITS) study similarly found no significant reduction in C. difficile infection rate [32]; however, this study was conducted in a hospital with low baseline antibiotic prescribing (it was ranked second-to-last in terms of antibiotic usage among its peer institutions), inherently limiting the risk of C. difficile infection among patients in the pre-ASP setting. In contrast to these findings, a meta-analysis specifically designed to assess the incidence of C. difficile infection relative to stewardship programs found a significantly reduced risk of infection based on 16 studies (RR 0.48, 95% CI 0.38–0.62, P < 0.001; I2 = 76%) [41], and the systematic review conducted by Filice et al [24] found a significant benefit with regard to the C. difficile infection rate in 4 of 6 studies. These results are consistent with those presented as evidence for the impact of stewardship on C. difficile infection by the CDC [42]. Aside from C. difficile infection, one retrospective observational study found that the 14-day reinfection rate (ie, reinfection with the same infection at the same anatomical location) was significantly reduced following stewardship intervention (0% vs. 10%, P = 0.009) [29]. This finding, combined with the C. difficile infection examples, provide evidence for better infection management of ASPs.
While the general trend seems to suggest mixed or no significant benefit for several clinical outcomes, it is important to note that variation in outcomes could be due to differences in the types of ASP interventions and intervention study periods across differing programs. Davey et al [37] found variation in prescribing outcomes based on whether restrictive (ie, restrict prescriber freedom with antimicrobials) or persuasive (ie, suggest changes to prescriber) interventions were used, and on the timeframe in which they were used. At one month into an ASP, restrictive interventions resulted in better prescribing practices relative to persuasive interventions based on 27 studies (effect size 32.0%, 95% CI 2.5%–61.4%), but by 6 months the 2 were not statistically different (effect size 10.1%, 95% CI –47.5% to 66.0%). At 12 and 24 months, persuasive interventions demonstrated greater effects on prescribing outcomes, but these were not significant. These findings provide evidence that different study timeframes can impact ASP practices differently (and these already vary widely in the literature). Considering the variety of ASP interventions employed across the different studies, these factors almost certainly impact the reported antimicrobial consumption rates and outcomes to different degrees as a consequence. A high degree of heterogeneity among an analyzed dataset could itself be the reason for net non-significance within single systematic reviews and meta-analyses.
Resistance
Another goal of ASPs is the prevention of antimicrobial resistance, an area where the evidence generally suggests benefit associated with ASP interventions. Resistance rates to common troublesome organisms, such as methicillin-resistant S. aureus (MRSA), imipenem-resistant P. aeruginosa, and extended-spectrum β-lactamase (ESBL)–producing Klebsiella spp were significantly reduced in a meta-analysis; ESBL-producing E. coli infections were not, however [22]. An ITS study found significantly reduced MRSA resistance, as well as reduced Pseudomonal resistance to imipenem-cilastin and levofloxacin (all P < 0.001), but no significant changes with respect to piperacillin/tazobactam, cefepime, or amikacin resistance [32]. This study also noted increased E. coli resistance to levofloxacin and ceftriaxone (both P < 0.001). No significant changes in resistance were noted for vancomycin-resistant enterococci. It may be a reasonable expectation that decreasing inappropriate antimicrobial use may decrease long-term antimicrobial resistance; but as most studies only span a few years, only the minute changes in resistance are understood [23]. Longer duration studies are needed to better understand resistance outcomes.
Of note is a phenomenon known as the “squeezing the balloon” effect. This can be associated with ASPs, potentially resulting in paradoxically increased resistance [43]. That is, when usage restrictions are placed on certain antibiotics, the use of other non-restricted antibiotics may increase, possibly leading to increased resistance of those non-restricted antibiotics [22] (“constraining one end [of a balloon] causes the other end to bulge … limiting the use of one class of compounds may be counteracted by corresponding changes in prescribing and drug resistance that are even more ominous” [43]). Karanika et al [22] took this phenomonen into consideration, and assessed restricted and non-restricted antimicrobial consumption separately. They found a reduction in consumption for both restricted and non-restricted antibiotics, which included “high potential resistance” antibiotics, specifically carbapenems and glycopeptides. In the study conducted by Cairns et al [28], a similar effect was observed; while the use of other classes of antibiotics decreased (eg, cephalosporins and aminoglycosides), the use of β–lactam–β–lactamase inhibitor combinations actually increased by 48% (change in use: +48.2% [95% CI 21.8%–47.9%]). Hohn et al [26] noted an increased usage rate of carbapenems, even though several other classes of antibiotics had reduced usage. Unfortunately, neither study reported resistance rates, so the impact of these findings is unknown. Finally, Jenkins et al [32] assessed trends in antimicrobial use as changes in rates of consumption. Among the various antibiotics assessed in this study, the rate of flouroquinolone use decreased both before and after the intervention period, although the rate of decreased usage slowed post-ASP (the change in rate post-ASP was +2.2% [95% CI 1.4%–3.1%], P < 0.001). They observed a small (but significant) increase in resistance of E. coli to levofloxacin pre- vs. post-intervention (11.0% vs. 13.9%, P < 0.001); in contrast, a significant decrease in resistance of P. aeruginosa was observed (30.5% vs. 21.4%, P < 0.001). While these examples help illustrate the concept of changes in antibiotic usage patterns associated with an ASP, at best they approximate the “squeezing the balloon” effect since these studies present data for antibiotics that were either restricted or for which restriction was not clearly specified. The “squeezing the balloon” effect is most relevant for the unintended, potentially increased usage of non-restricted drugs secondary to ASP restrictions. Higher resistance rates among certain drug classes observed in the context of this effect would constitute a drawback to an ASP program.
Adverse Effects
Reduced toxicities and adverse effects are expected with reduced usage of antimicrobials. The systematic review conducted by Filice et al [24] examined the incidence of adverse effects related to antibiotic usage, and their findings suggest, at the least, that stewardship programs generally do not cause harm, as only 2 of the studies they examined reported adverse events. Following stewardship interventions, 5.5% of the patients deteriorated; and of those, the large majority (75%) deteriorated due to progression of oncological malignancies. To further illustrate the effect of stewardship interventions on toxicities and side effects of antimicrobials, Schuts et al demonstrated that the risk of nephrotoxicity while on antimicrobial therapy was reduced based on 14 studies of moderate heterogeneity as a result of an ASP (OR 0.46, 95% CI 0.28–0.77, P = 0.003; I2 = 34%) [39,44]. It is intuitive that reduced drug exposure results in reduced adverse effects, as such these results are expected.
Economic Outcomes
Although the focus of ASPs is often to improve clinical outcomes, economic outcomes are an important component of ASPs; these programs bring associated economic value that should be highlighted and further detailed [22,45,46]. Since clinical outcomes are often the main objective of ASPs, most available studies have been clinical effect studies (rather than economic analyses), in which economic assessments are often a secondary consideration, if included.
As a result, cost evaluations are conducted on direct cost reductions whereas indirect cost reductions are often not critically evaluated. ASPs reduce hospital expenditures by limiting hospital-acquired infections and the associated medical costs where they are effective at decreasing consumption of antimicrobials [22,45], and by reducing antibiotic misuse, iatrogenic infections, and the rates of antibiotic-resistant organisms [47]. In one retrospective observational study, annual costs of antibiotics dropped by 33% with re-implementation of an ASP, mirrored by an overall decrease in antibiotic consumption of about 10%, over the course of the intervention study period [30]. Of note is that at 1 year post-ASP re-implementation, antibiotic consumption actually increased (by 5.4%); however, because antibiotic usage had changed to more appropriate and cost-effective therapies, cost expenditures associated with antibiotics were still reduced by 13% for that year relative to pre-ASP re-implementation. Aside from economic evaluations centered on consumption rates, there is the potential to further evaluate economic benefits associated with stewardship when looking at other outcomes, including hospital LOS [22], as well as indirect costs such as morbidity and mortality, societal, and operational costs [46]. Currently, these detailed analyses are lacking. In conjunction with more standardized clinical metrics, these assessments are needed to better delineate the full cost effectiveness of ASPs.
