Introduction
One key aim of antimicrobial stewardship programs (ASPs) is to limit the breadth of the spectrum of antimicrobial activity to reduce selective pressure for antimicrobial resistance. Reference Barlam, Cosgrove and Abbo1 Although no standard definition exists for broad-spectrum antibiotic therapy, antimicrobials with activity against clinically relevant pathogens and human microbiota beyond the target pathogen may be associated with a greater risk of antimicrobial resistance and C. difficile infections. Reference Nakai, Hagihara and Kato2–Reference Coburn, Daneman, MacFadden and Rooney4
There is wide variability in antibiotic prescribing across hospitals in terms of both the volume and type of antibiotics prescribed. For example, a study from Ontario, Canada, found a 7.4-fold difference between the hospital with the lowest use and the hospital with the highest use. The variation in use persisted even after adjusting for population characteristics that predict antibiotic use. This suggests an opportunity to reduce practice variation and improve antibiotic prescribing to reduce the risk of harm to hospitalized patients. Reference Zanichelli, Monnier and Gyssens5–Reference Goodman, Baghdadi and Magder7 Although hospital ASPs have been widely implemented to improve the quality of antimicrobial prescribing, the extent of program implementation, particularly regarding the level of resourcing and funding for program operation, varies substantially across hospitals. Reference Leung, Wu, Langford and Garber8,Reference Leung, Quirk, Muir, Daneman, Schwartz and Langford9
We have previously shown that the presence of designated funding and resources for ASPs is associated with a 13% lower risk-adjusted volume of antimicrobial use (AMU) compared to programs without these resources. Reference Langford, Wu and Brown10 However, to our knowledge, the impact of ASP resourcing on hospital-wide broad-spectrum antibiotic use has not been previously evaluated. Our objective was to investigate whether the presence of designated funding and resources for ASPs is associated with a lower proportion of broad-spectrum antibiotic use in hospitals
Methods
General study design
We conducted a cross-sectional study on AMU among Ontario hospitals as part of the Ontario Hospital ASP Landscape Survey. Details regarding this periodic survey are provided elsewhere. Reference Leung, Quirk, Muir, Daneman, Schwartz and Langford9 Briefly, this voluntary online survey is conducted by Public Health Ontario every 2–3 years and collects data about program structure (including designated funding and/or resources), activities, and AMU. The relevant question from this survey asks “Are there designated funding/resources for your ASP?”. This question was designed to capture the presence of allocated resources to their stewardship program whether or not there was a specific budget allocation for these resources. The survey was available from 3 October 2023 to 29 December 2023 and was disseminated via targeted e-mails to all acute care, inpatient rehabilitation and complex continuing care (CCC) hospitals across the province. There was no incentive offered for participation; however, there were multiple e-mail and telephone reminders to encourage response. All AMU data were self-reported by hospitals using a standardized spreadsheet.
Eligibility
Hospitals were included if they provided at least six months of complete AMU data. Hospitals were excluded if they did not provide data on any of the broad-spectrum antibiotics or the total antibacterial AMU data. Specific data points (eg, AMU data of an antibiotic during a specific fiscal period) were excluded if they were implausible (eg the use data of a specific antibiotic is higher than total antibiotic use data).
Data on antibiotic utilization
Hospital ASPs provided total and class-specific AMU aggregated across the fiscal years from April 2020 to March 2023 as either defined daily dose (DDD or days of therapy (DOT) for all systemic antibacterials (World Health Organization (WHO) Anatomical Therapeutic Chemical Classification J01) 11 administered via the enteral or parenteral route as well as the following selected classes of antibiotics, which for the purposes of this study were defined as broad spectrum: fluoroquinolones (ciprofloxacin, levofloxacin, and moxifloxacin), third-generation cephalosporins (ceftriaxone, cefotaxime, and ceftazidime), beta-lactam/beta-lactamase inhibitor combinations (amoxicillin-clavulanate and piperacillin-tazobactam), carbapenems (ertapenem, meropenem, and imipenem-cilastatin), clindamycin, and parenteral vancomycin. These antibiotics were selected due to their broad spectrum of antimicrobial activity and their potential to increase the risk of C. difficile infection. Additionally, most of these antibiotics are classified in the “Watch” category in the WHO Access, Watch, Reserve (AWaRe) list. 12 Hence, careful monitoring of these antibiotics is required to ensure they are used responsibly, preserving their effectiveness and minimizing the emergence and spread of resistant infections.
These antibiotics were further categorized as total broad-spectrum antibiotics (all the antibiotic classes identified above) and anti-pseudomonal antibiotics (ciprofloxacin, levofloxacin, ceftazidime, piperacillin-tazobactam, meropenem, and imipenem-cilastatin).
Survey respondents from corporations with more than one hospital facility provided AMU data for each facility within the organization. If a hospital provided both DOT and DDD data, only the DOT data was used for analysis. The participating hospitals provided the number of patient days (PDs) from inpatient admissions separately as a measure to estimate the volume of patient admissions in the facility.
Exposure
The exposure of interest was ASP funding status which was defined as whether or not hospitals reported having designated funding and/or resources for their program.
Outcomes
The primary outcome was the proportion of all antibiotic use that is broad-spectrum which was calculated by dividing the volume of broad-spectrum AMU (DDD or DOT) by each hospital’s total volume of AMU (DDD or DOT) including all J01 antibacterial agents. The calculated proportions were stratified by hospital type (community, teaching, or CCC/rehabilitation) and regions (Toronto, Central, West, East, North-West, and North-East) as defined by the Ontario Ministry of Health. 13 Secondary outcomes include the proportion of AMU for each of the antibiotic classes described above.
Statistical analysis
Descriptive statistics illustrating the median, range, and interquartile range were calculated for the proportion of each category of antibiotics. These data were stratified by hospital type and region to account for potential differences in patient populations and epidemiology across different hospitals.
To evaluate the relationship between funding status and the proportion of broad-spectrum antibiotic use, we constructed logistic mixed models. The primary explanatory variable in our models was ASP funding status. The outcome was each hospitals proportion of broad-spectrum antibiotic use and a weight for the total volume of AMU. The fixed effect in the models was funding status, while the random effect was the hospital, which accounts for clustering of the outcome within hospitals. Reference Harrison14 This approach enabled us to quantify the extent to which funding status influenced prescribing practices, with the results expressed as odds ratios (OR) with 95% confidence intervals (CI). An adjusted analysis was performed which included fixed effects for hospital type and metric type (DDD vs DOT) in the model. To further explore the association between funding/resourcing status and specific classes of antibiotic use independent of the total volume of use, we performed a sensitivity analysis using the denominator of only broad-spectrum antibiotic use. All statistical analysis and forest plots were generated using R statistical software version 4.4.0 (R Foundation, Vienna, Austria).
Results
The overall survey response rate was 70% (90/129) of hospital corporations in Ontario and 39% of hospitals provided AMU data (79/202). After exclusions, the final cohort included 63 unique sites, representing 31% (63/202) of individual hospitals in the province. This included 9 CCC/rehabilitation, 16 teaching, 15 large community, 15 medium community, and 8 small community hospitals. Representation was predominantly from the southern, more populated regions of the province. Of the 63 hospitals, 48 (76%) reported designated ASP funding or resources. Funding/resourcing was more commonly reported in acute teaching hospitals (n = 15, 94%) compared to community hospitals (n = 26, 68%) and CCC and rehabilitation (n = 7, 78%; Table 1).
Hospital participation and presence of ASP funding/resources
Overall antibiotic utilization
The median total AMU for hospitals reporting DDD per 1000 PDs was 413 (IQR 275–494; range 62–690) (n = 12), and for hospitals reporting DOT, it was 415 (IQR 305–529; range 109 to 1656) (n = 51).
Broad-spectrum antibiotic utilization
The median proportion of total broad-spectrum AMU for hospitals was 52.5% (IQR 48.4%–59.7%). Anti-pseudomonal agents (median 22.8%, IQR 18.4%–25.8%), third-generation cephalosporins (median 16.3%, IQR 13.4%–21.3%), and beta-lactam/beta-lactamase inhibitors (median 15.6%, IQR: 11.7%–19.0%) were the most commonly represented categories of broad-spectrum antibiotic use (Table 2).
Antimicrobial stewardship program resourcing and impact on broad-spectrum antibiotic use
*Anti-pseudomonal agents include ciprofloxacin, levofloxacin, ceftazidime, piperacillin-tazobactam, meropenem, and imipenem-cilastatin.
Funding status and broad-spectrum antibiotic use
After adjusting for hospital type and metric type, the presence of ASP funding/resources was not associated with overall broad-spectrum antibiotic use (OR 1.02, 95% CI: 0.83–1.28). However, funding/resourcing was associated with a lower proportion of fluoroquinolones (OR 0.67, 95% CI: 0.46–0.96, P = 0.03), clindamycin (OR 0.69, 95% CI: 0.47–1.00, P < 0.05), and anti-pseudomonal agents (OR 0.76, 95%CI: 0.59–0.98, P = 0.03). Funding/resourcing status was not associated with higher broad-spectrum antibiotic use for any of the classes studied. Table 2 includes the unadjusted and adjusted OR illustrating the association between ASP funding/resourcing and broad-spectrum antibiotic use. The association between funding/resourcing status and lower use of fluoroquinolones, anti-pseudomonal agents, and clindamycin persisted when using the denominator of only total broad-spectrum antibiotic use (Supplemental Table 1).
Discussion
In this province-wide cross-sectional study, we identified variability in broad-spectrum antibiotic prescribing across 63 hospitals, and ASP funding status was not associated with lower total broad-spectrum antibiotic use. However, hospitals with funding and resources for their ASP used lower proportions of fluoroquinolones, anti-pseudomonal agents, and clindamycin.
The presence of designated funding and/or resources for hospital ASPs has previously been associated with a lower volume of risk-adjusted antibiotic prescribing. Reference Langford, Wu and Brown10 This study builds upon these findings to suggest that funding status may also affect antibiotic selection. The US Centers for Disease Control and Prevention (CDC) highlights the importance of a hospital leadership commitment to ensuring ASPs have adequate human, financial, and technology-related resourcing. 15 Several national recommendations exist to provide targets for optimal staffing of ASPs; Reference Greene, Nesbitt and Nelson16 however, there is evidence these staffing targets are largely unmet. Reference Leung, Quirk, Muir, Daneman, Schwartz and Langford9
In our study, the most notable association of funding status with prescribing relates to fluoroquinolones, clindamycin, and anti-pseudomonal agents. Increasing rates of antimicrobial resistance, several black-box warnings, and an elevated risk of C. difficile infection have led ASPs to implement interventions to avoid unnecessary fluoroquinolone use. Reference Claeys, Hopkins, Vega and Heil17–Reference Langford, Seah, Chan, Downing, Johnstone and Matukas19 Similarly, clindamycin has exhibited one of the highest odds of inciting C. difficile infection, yet is rarely a first-line agent for common infections managed in the hospital. Reference Miller, Arakkal and Sewell20 As such, this agent is often used inappropriate and is a potential target for ASPs. Reference Lawes, Lopez-Lozano and Nebot21,Reference Feihl, Querbach and Hapfelmeier22 Our study suggests that targeted initiatives may be more successful if the ASP has designated resources, allowing adequate staffing and time to make impactful changes on prescribing. Reference López-Cortés, Delgado-Valverde and Moreno-Mellado23,Reference Tabah, Cotta and Garnacho-Montero24
Strengths of this study include a relatively large sample representing various hospital types from across the province. Limitations include the observational nature of these data, which precludes a definitive assessment of causality. However, unpublished data from our team suggest minimal changes in ASP staffing over the study period, limiting the possibility of reverse causation. We lacked patient-level data to adjust for various confounders linked to broad-spectrum prescribing (eg, patient comorbidities and acuity); however, we used hospital type as a proxy for these factors. Additionally, most of the hospitals included represent community and chronic care facilities which incorporate more homogenous patient populations. Although we did not directly assess antibiotic appropriateness, several of the broad-spectrum classes of antibiotics used represent first-line agents for common community and hospital-acquired infections (eg, ceftriaxone and piperacillin-tazobactam, respectively). As such, the usage of such classes of antibiotics may be largely aligned with clinical practice guidelines and less likely to be affected by the intensity of ASP activities. This observation is consistent with our findings—that the usage of agents that represent first-line therapy (eg, third-generation cephalosporins and beta-lactam beta-lactamase inhibitors) for common infections was not associated with the presence of ASP resources. Although there is no widely accepted definition or metric for “broad spectrum,” some researchers have devised spectrum scores largely based on clinical spectrum of activity. Reference Madaras-Kelly, Jones, Remington, Hill, Huttner and Samore25,Reference Gerber, Hersh, Kronman, Newland, Ross and Metjian26 Future efforts are needed to better define “spectrum” and classify antibiotics based on not only their clinical spectrum, but also their propensity for selecting antimicrobial resistance. Only about one-third of hospitals in Ontario are able to prepare and submit AMU data. This may have introduced some degree of selection bias. The data included in this study may not be representative of non-participating facilities, many of which are lacking in ASP funding and resources. As such, these findings may underestimate the impact of ASP resourcing on AMU, given that all the programs able to participate had at least sufficient resourcing to prepare and submit their AMU data. Finally, the data collection period overlaps with the COVID-19 pandemic. The pandemic affected ASP activities Reference Leung, Quirk, Muir, Daneman, Schwartz and Langford9 and antimicrobial prescribing Reference Elligsen, Wan and Lam27 . Ongoing efforts are needed to quantify the impact of ASP resourcing beyond the pandemic period.
In light of the threat of antimicrobial resistance (AMR), the extent of ASP resourcing may be a critically important factor in mitigating the further emergence and spread of these drug-resistant organisms. Future prospective research evaluating various funding models in different healthcare settings may shed light on optimal approaches to leverage resources for effective ASPs. Cost-effectiveness studies that comprehensively assess cost savings of ASPs in terms of avoidance of negative outcomes may provide robust data needed to convince decision-makers of the importance of adequate staffing.
Conclusion
Hospital ASPs with designated funding and resources are associated with reduced proportional use of fluoroquinolones, clindamycin, and anti-pseudomonal agents but not the overall proportion of broad-spectrum prescribing. The escalating prevalence of AMR necessitates efforts to improve resource allocation and evaluate the impact of such resourcing on hospital ASPs.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/ash.2024.461.
Acknowledgments
This work was carried out as part of routine operations. No additional funding was received. We wish to thank antimicrobial stewards across Ontario for their participation in the Ontario ASP Landscape Survey.
Financial support
None.
Competing interests
None.
Open access
