It has been postulated that the coronavirus disease 2019 (COVID-19) pandemic has disrupted the functions of antimicrobial stewardship programs (ASPs), which are considered one of the key tools for mitigating the evolution of multidrug-resistant organisms (MDROs). Reference Stevens, Patel and Nori1,Reference Stevens, Doll, Pryor, Godbout, Cooper and Bearman2 This concern has arisen since members of the ASPs have been diverted in large part to develop institutional guidance and policies for the management of COVID-19. Reference Furukawa and Graber3 Moreover, increases in empiric antibiotic prescribing during the COVID-19 pandemic have been described. Reference Langford, So and Raybardhan4 However, the extent to which long-term trends in antimicrobial use across different antimicrobial stewardship program (ASP) models have changed throughout the COVID-19 pandemic is less well understood. We have described temporal fluctuations and longitudinal trends in antimicrobial prescribing of key agents across an integrated healthcare system with different ASP models.
Methods
This study was conducted at 12 hospitals that are part of an integrated health system in Iowa. Data for antibiotic days of therapy (DOT) and days present were extracted from a centralized database. Only medical–surgical and intensive care units (ICUs) were included. The antibiotics most frequently prescribed at our facilities were selected for analysis: meropenem, piperacillin–tazobactam, cefepime, ceftriaxone, vancomycin, azithromycin, doxycycline, and levofloxacin. None of the antibiotics included are subject to preauthorization. We collected data from January 1, 2019, to February 28, 2021, which encompasses a period prior to the widespread availability of vaccines and preceding the circulation of the δ (delta) and δ (omicron) variants. The prepandemic period was defined as January 1, 2019, to February 29, 2020, and the pandemic period was defined as March 1, 2020, to February 28, 2021. During the first year of the pandemic, the state of Iowa experienced 2 significant increases in community spread with commensurate increase in hospitalizations for COVID-19. The first peak occurred in early May 2020 (peak number of patients hospitalized, 417) and the second peak occurred in mid-November 2020 (peak number of patients hospitalized, 1,510). 5
Antimicrobial stewardship program models
The daily antimicrobial stewardship activities and the composition of the staff performing them varied by site. Stewardship activities remained unchanged throughout the study period. The characteristics of the hospitals supported by each ASP model, as well as the ASP members and workflow are described in Table 1.
Table 1. Hospital and Antimicrobial Stewardship Program (ASP) Model Characteristics
Note. ICU, intensive care unit; ID, infectious disease.
Statistical analysis
We examined the trends in antibiotic DOT per 1,000 days present among the 3 different ASP models. We used joinpoint regression to determine the number of joinpoints, the monthly percentage changes (MPCs), and the average monthly percent changes (AMPCs). 6 The joinpoints are points at which the trend changes. The MPC characterizes changes in trends in antibiotic DOT rates occurring at any point during the entire observation period. The AMPC summarizes the trend over prespecified fixed intervals, which in our study were the prepandemic period (January 2019–February 2020) and the pandemic period (March 2020–February 2021). Models were fit to log-transformed antibiotic DOT rates, and permutation analysis was used to select the best-fit model. An autocorrelated error structure was selected to account for autoregression in prescribing rates over time. Analyses were conducted on the Joinpoint Regression Program version 4.7 software (National Cancer Institute, Bethesda, MD).
Results
Facilities using antimicrobial stewardship model A
In these facilities, estimates of the MPC for the entire observation showed significant changes in trend of use of piperacillin–tazobactam, ceftriaxone, azithromycin and doxycycline (Table 2, Supplementary Fig. 1B, 1D, 1F, and 1G). During the prepandemic period, AMPC estimates showed monthly increases in the use of ceftriaxone, vancomycin, azithromycin, and a decrease in levofloxacin use (Supplementary Table 1). During the pandemic period, the AMPC showed monthly increases in the use of cefepime, piperacillin–tazobactam, and vancomycin. The AMPC of vancomycin was the same for both the prepandemic and pandemic periods (+0.4%; 95% CI, 0.1–07; P = .04), reflecting a longstanding trend.
Table 2. Monthly Percent Change in Antibiotic Days of Therapy Per 1,000 Days Present According to Antimicrobial Stewardship Program (ASP) Model, January 1, 2019, to February 28, 2021
Facility using antimicrobial stewardship model B
At this site, estimates of the MPC for the entire observation period detected significant changes in the trends in cefepime and azithromycin use (Table 2 and Supplementary Fig. 2C and 2F). For the prepandemic period, AMPC estimates showed monthly increases in the use of meropenem, cefepime, ceftriaxone, vancomycin, and azithromycin, and a decrease in levofloxacin use (Supplementary Table 1). For the pandemic period, AMPCs showed monthly increases in the use of meropenem, ceftriaxone and vancomycin, and a decrease in levofloxacin use (Supplementary Table 1). The AMPCs for meropenem (+2.3; 95% CI, 0.8–3.8; P < .01), ceftriaxone (+1.0; 95% CI, 0.2–1.7; P = .02), vancomycin (+2.4; 95% CI, 1.0–3.5; P < .01), and levofloxacin (−2.4; 95% CI, −3.3 to −1.5; P < .01) remained the same for the prepandemic and pandemic periods, reflecting longstanding trends.
Facilities using antimicrobial stewardship model C
In these facilities, the MPC estimates for the entire observation period indicated multiple significant fluctuations in the use of meropenem, cefepime, azithromycin, doxycycline, and levofloxacin (Table 2 and Supplementary Figs 3A, 3C, 3F, 3G and 3H). For the prepandemic period, the AMPCs showed monthly increases in the use of meropenem, piperacillin–tazobactam, ceftriaxone, and doxycycline, and a decreases in the use of vancomycin and levofloxacin (Supplementary Table 1). For the pandemic period, the AMPC showed monthly increases in the use of piperacillin–tazobactam and ceftriaxone and a decrease in vancomycin use (Table 2). The AMPCs of piperacillin–tazobactam (+0.6; 95% CI, 0.2–1; P < .01), ceftriaxone (+0.9; 95% CI, 0.4–1.3; P < .01), and vancomycin (−0.3; 95% CI, −0.5 to −0.1; P = .05) remained the same for both the prepandemic and pandemic periods, reflecting the longstanding trend.
Discussion
Across hospitals using different ASP models, we identified multiple fluctuations in the rates of antibiotic use throughout the study period. In most cases, the average monthly percent changes reflected trends that preceded the COVID-19 pandemic.
Up to 75% of patients with COVID-19 are prescribed antibiotics, and rates of prescribing have decreased throughout the pandemic. Reference Langford, So and Raybardhan4 Our assessment of longitudinal trends in prescribing revealed fluctuations that, in most instances, did not reach a statistically significant deviation from the existing trend. There is a need for development of ASPs in settings with both limited access to the expertise of infectious diseases specialists, particularly during the COVID-19 pandemic, as well as a lack of reports of stewardship practices. Reference Pierce and Stevens7 Our study contributes to the literature on this topic by describing 3 ASP models in urban and semirural areas and by describing trends of use of key antibiotic agents under different stewardship practices.
The limitations of our study include use of registry-type data obtained from a centralized database in which accuracy depends on appropriate capture of data in the medication administration record. This aspect is mitigated by previous reviews on the accuracy of this data performed by clinical pharmacists. Also, we did not have data for the precise indication of antibiotic prescribing, which precluded us from evaluating changes in trends used specifically for respiratory infection. Furthermore, we did not assess the role of individual tools for mitigating antimicrobial use such as procalcitonin trends. Reference Martin, Philbin, Hughes, Bergin and Talento8
In conclusion, across 3 different ASP models, the core stewardship activities were maintained during the COVID-19 pandemic. Changes in antibiotic use were limited to temporal fluctuations and, for most agents, longstanding trends continued. Continued support and development of ASPs in accordance with local resources is crucial to their success.
Supplementary material
For supplementary material accompanying this paper visit https://doi.org/10.1017/ash.2022.39
Acknowledgments
Financial support
No financial support was provided relevant to this article.
Conflicts of interest
All authors report no conflicts of interest relevant to this article.
Open access
