Days of Antibiotic Spectrum Coverage: A Novel Metric for Inpatient Antibiotic Consumption 

Abstract

Background

Days of therapy (DOT), the most widely used benchmarking metric for antibiotic consumption, may not fully measure stewardship efforts to promote use of narrow-spectrum agents and may inadvertently discourage the use of combination regimens when single-agent alternatives have greater adverse effects. To overcome the limitations of DOT, we developed a novel metric, days of antibiotic spectrum coverage (DASC), and compared hospital performances using this novel metric with DOT.

Methods

We evaluated 77 antibiotics in 16 categories of antibacterial activity to develop our spectrum scoring system. DASC was then calculated as cumulative daily antibiotic spectrum coverage (ASC) scores. To compare hospital benchmarking using DOT and DASC, we conducted a retrospective cohort study of adult patients admitted to acute care units within the Veterans Health Administration system in 2018. Antibiotic administration data were aggregated to calculate each hospital’s DOT and DASC per 1000 days present (DP) for ranking.

Results

The ASC score for each antibiotic ranged from 2 to 15. There was little correlation between DOT per 1000 DP and DASC per DOT, indicating that lower antibiotic consumption at a hospital does not necessarily mean more frequent use of narrow-spectrum antibiotics. The differences in each hospital’s ranking between DOT and DASC per 1000 DP ranged from −29.0% to 25.0%, respectively, with 27 hospitals (21.8%) having differences >10%.

Conclusions

We propose a novel composite metric for antibiotic stewardship, DASC, that combines consumption and spectrum as a potential replacement for DOT. Further studies are needed to evaluate whether benchmarking using the DASC will improve evaluations of stewardship.

antibiotic consumption metric, antibiotic spectrum, antibiotic stewardship, hospital benchmarking

Days of therapy (DOT), often normalized by the days of inpatient stay, is the most widely used metric used to quantify antibiotic consumption in inpatient settings [1, 2]. DOT reflects the total volume of antibiotic consumption in the inpatient setting and comes with very important advantages, such as objectivity and scalability.

However, DOT has 3 major limitations, especially when used as a benchmarking metric for hospital comparisons. First, antibiotic consumption can be heavily affected by case-mix (the diversity, complexity, and severity of patient illnesses) and the level of care provided in each care unit [3, 4]. To address this issue, the Center for Diseases Control and Prevention introduced the Standardized Antimicrobial Administration Ratio (SAAR) in the National Healthcare Safety Network (NHSN) Antimicrobial Use and Resistance (AUR) module [2]. SAAR is based on DOT but is risk adjusted for hospital-level and unit-level factors [5]. Second, DOT simply counts the days a patient received antibiotics, not accounting for the spectrum or types of microorganisms targeted. This can be a serious problem when comparing monotherapy and combination regimens that provide a similar spectrum of activity [6]. For example, piperacillin-tazobactam monotherapy and a combination of cefepime plus metronidazole have a similar spectrum of activity, but the latter would have twice as many DOTs compared with the former if given for an equal duration. Third, DOT does not always reflect efforts to avoid the use of unnecessary broad-spectrum agents through judicious empiric therapy or de-escalation. For example, when cefepime is chosen over meropenem as a narrower empiric therapy or for subsequent de-escalation, DOT is the same despite a clear stewardship benefit to decreasing the risk of infection or colonization with carbapenem-resistant Enterobacteriaceae and unnecessary anaerobic coverage [7, 8]. The second and third limitations cannot be addressed by risk adjustments, as long as the metric is based on conventional DOT.

The antibiotic spectrum index (ASI), a scoring system for antibiotic spectrums, was proposed by Gerber et al [9]. The ASI assigns 1 point for each organism category that an antibiotic covers (total, 14 categories); the sum of points determines the index for that antibiotic agent. The ASI has several important advantages, including simplicity, ease of operationalization, and reflection of clinically important spectrums. However, it also has a few limitations. First, the selection criteria and definitions of categories for scoring are not clearly defined. Second, it does not comprehensively describe all antibiotic agents used in clinical practice in the United States and globally. Although previous studies have used ASI to calculate the average antibiotic spectrum by specific conditions and applied it to examine the impact of antibiotic stewardship program (ASP) interventions as an instantaneous evaluation of antibiotic spectrum coverage [10–12], it has not been used as a cumulative measure for the purposes of hospital or unit benchmarking.

Inspired by the ASI, we developed a novel antibiotic consumption metric, days of antibiotic spectrum coverage, that reflects both quantity and spectrums of antibiotic agents to allow more accurate benchmarking, which can be used as a replacement of DOT. The aims of the current study are to describe the creation process of a novel metric and to demonstrate the difference between this novel metric and a conventional metric (DOT) when used for hospital benchmarking.

METHODS

Assessments for Antibiotic Spectrums and Development of DASC

To develop an antibiotic spectrum scoring system, the antibiotic spectrum coverage (ASC) score, we configured categories of clinically important bacterial pathogens into 2 domains, wild-type organisms without acquired resistance and commonly isolated microorganisms with specific mechanisms of acquired resistance. The wild-type category includes 11 organism groups, including methicillin-susceptible Staphylococcus aureus, Streptococcus species (groups A and B Streptococcus spp.), Enterococcus faecalis, oral anaerobes (Peptostreptococcus/Fusobacterium/Veillonella spp.), Bacteroides fragilis, Moraxella catarrhalis/Hemophilus influenzae, Escherichia coli/Klebsiella pneumoniae, Enterobacter/Serratia/Citrobacter spp. (as representations of AmpC-producing organisms), Pseudomonas aeruginosa, Acinetobacter baumannii, and atypical organisms (Rickettsia/Chlamydophila/Mycoplasma/Legionella spp.).

The acquired resistance category includes 5 organism groups, including extended-spectrum β-lactamase–producing Enterobacteriaceae, methicillin-resistant S. aureus (β-lactam resistance mediated by mecA), penicillin-resistant Streptococcus pneumoniae (decreased susceptibility to penicillin induced by mutations of penicillin-binding proteins), vancomycin-resistant Enterococcus spp. (vancomycin resistance mediated by vanA or vanB), and carbapenem-resistant Enterobacteriaceae (carbapenem resistance mediated by class A carbapenemases, such as K. pneumoniae carbapenemase). We then selected 77 antibiotic agents currently marketed in the United States [2] and additional agents are available internationally.

We then evaluated activities of 77 agents in 16 organism categories (11 wild type and 5 acquired resistance), through literature searches conducted using PubMed and EmBase, as well as widely accepted textbooks [13, 14], and assigned dichotomous classifications to each antibiotic-organism category combination. If no explicit statement was found in the existing literature or there was a discrepancy between in vitro activity and clinical recommendations, we convened an expert panel, including infectious diseases specialists, microbiologists, and pharmacists, to form a consensus. ASC scores were assessed to evaluate in vitro activities of agents against naive organisms and the potential to induce resistance of organisms through exposure, and not intended to provide therapeutic guidance. Therefore, a wild-type organism that is usually susceptible to an agent in vitro but may develop resistance in vivo quickly after exposure still receives a score of 1 (eg, third-generation cephalosporins and AmpC-producing Enterobacteriaceae). For simplicity and operationalizability, the ASC score does not consider antibiotic pharmacokinetics, pharmacodynamics, or toxicity.

The DASC is calculated by aggregating the ASC scores of all antibiotics used in a day, including multiple antibiotics with the same spectrum. For example, ceftriaxone’s spectrum of coverage includes S. aureus, Streptococcus species, M. catarrhalis/H. influenzae, E. coli/K. pneumoniae, Enterobacter/Serratia/Citrobacter spp., and penicillin-resistant S. pneumoniae (ASC score, 6), and azithromycin includes S. aureus, Streptococcus species, oral anaerobes, M. catarrhalis/H. influenzae, E. coli/K. pneumoniae, atypical organisms (ASC score, 6). Therefore, if a patient uses ceftriaxone and azithromycin, the DASC would be 12 per day, and the DOT would be 2 per day.

Study Population for Comparison Between DOT and DASC

To compare hospital benchmarking results between DOT and DASC, we conducted a retrospective cohort study of all patients admitted to acute care units within the Veterans Health Administration (VHA) between 1 January and 31 December 2018. These 124 hospitals were located in all 48 continental states, the District of Columbia, and Puerto Rico.

We focused on the comparison between DOT and DASC for antibiotic agents currently included in the antimicrobial use option of the NHSN [2]; thus, agents excluded from analysis included cefiderocol, colistin (colistimethate), delafloxacin, eravacycline, imipenem-cilastatin-relebactam, lefamulin, meropenem-vaborbactam, omadacycline, piperacillin, plazomicin, and ticarcillin-clavulanate.

For all adult inpatients in each hospital, we collected (1) days present (DP) for inpatients aged ≥18 years, (2) antibiotics administered, (3) DOT and DASC for each patient, and (4) patient location (medical vs surgical and intensive care unit [ICU] vs non-ICU). DP and DOT were calculated according to the method described in the NHSN AUR module [2]. We ranked each hospital based on antibiotic use evaluated by DOT and DASC per 1000 DP and defined higher ranking as lower antibiotic use in DOT per 1000 DP or a narrower spectrum in DASC per 1000 DP. We defined an operationally impactful ranking change a priori as ≥10%.

We also calculated the DASC per DOT in the study period and evaluated the relationship between DASC per DOT and DOT per 1000 DP. DASC per DOT represents an average ASC score per patient per treatment day in each hospital. Stratified analyses for ICU versus non-ICU and medical versus surgical units were also performed. All data were obtained from the VHA Corporate Data Warehouse. Inpatient antibiotic use was captured by a barcoding medication administration system.

Statistical Analysis

The correlation between DOT per 1000 DP and DASC per DOT and the ranking of DOT and DASC per 1000 DP were assessed using Spearman rank-order correlation test. Statistical analyses were performed with R software, version 4.0.3 (R Foundation for Statistical Computing). Antibiotic consumption data were collected as a part of an operational quality improvement project and exempted from institutional review board approval. For the purpose of this report beyond the operational needs, the Institutional Review Board of the University of Iowa and the Research and Development Committee at Iowa City Veterans Health Care System approved this study with a waiver of informed consent.

RESULTS

Assignments of ASC Scores to Antibiotic Agents

ASC scores for the 77 evaluated antibiotics ranged from 2 to 15, with a median of 6 (interquartile range, 5–9). ASC score assignments for antibiotic agents in each antibiotic class are shown in Table 1. The supporting evidence for ASC scores and the antibiotic-organism combinations that required expert panel discussions are shown in Supplementary Table 1.

Table 1.

Antibiotic Spectrum Coverage Score

Antibiotic Class	Antibiotics	ASC Scorea	Spectrum by Microorganism Category
			Wild Type											Acquired Resistance
			S. aureus	Streptococcus spp.	E. faecalis	Anaerobes (Oral)	B. fragilis	Moraxella/H. influenzae	E. coli/K. pneumoniae	Enterobacter/Serratia/Citrobacter	P. aeruginosa	A. baumannii	Atypical	ESBL	MRSA	PRSP	VRE	CRE
Aminoglycoside	Amikacin/amikacin liposomal	8	1	0	0	0	0	1	1	1	1	1	0	1	1	0	0	0
	Gentamicin	9	1	0	0	0	0	1	1	1	1	1	0	1	1	0	0	1
	Plazomicin	9	1	0	0	0	0	1	1	1	1	1	0	1	1	0	0	1
	Tobramycin	9	1	0	0	0	0	1	1	1	1	1	0	1	1	0	0	1
β-Lacta/β-lactamase inhibitor combination	Amoxicillin-clavulanate	7	1	1	1	1	1	1	1	0	0	0	0	0	0	0	0	0
	Ampicillin-sulbactam	8	1	1	1	1	1	1	1	0	0	1	0	0	0	0	0	0
	Ceftazidime-avibactam	8	0	1	0	0	0	1	1	1	1	1	0	1	0	0	0	1
	Ceftolozane-tazobactam	8	0	1	0	1	1	1	1	1	1	0	0	1	0	0	0	0
	Ticarcillin-clavulanate	10	1	1	0	1	1	1	1	1	1	1	0	1	0	0	0	0
	Piperacillin-tazobactam	11	1	1	1	1	1	1	1	1	1	1	0	1	0	0	0	0
	Imipenem-cilastatin-relebactam	13	1	1	1	1	1	1	1	1	1	1	0	1	0	1	0	1
	Meropenem-vaborbactam	13	1	1	1	1	1	1	1	1	1	1	0	1	0	1	0	1
Carbapenem	Ertapenem	9	1	1	0	1	1	1	1	1	0	0	0	1	0	1	0	0
	Doripenem	12	1	1	1	1	1	1	1	1	1	1	0	1	0	1	0	0
	Imipenem-cilastatin	12	1	1	1	1	1	1	1	1	1	1	0	1	0	1	0	0
	Meropenem	12	1	1	1	1	1	1	1	1	1	1	0	1	0	1	0	0
Cephalosporin	Cefadroxil	3	1	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0
	Cefazolin	3	1	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0
	Cefixime	3	0	1	0	0	0	1	1	0	0	0	0	0	0	0	0	0
	Cephalexin	3	1	1	0	0	0	0	1	0	0	0	0	0	0	0	0	0
	Cefaclor	4	1	1	0	0	0	1	1	0	0	0	0	0	0	0	0	0
	Cefdinir	4	1	1	0	0	0	1	1	0	0	0	0	0	0	0	0	0
	Cefpodoxime	4	1	1	0	0	0	1	1	0	0	0	0	0	0	0	0	0
	Cefprozil	4	1	1	0	0	0	1	1	0	0	0	0	0	0	0	0	0
	Cefuroxime	4	1	1	0	0	0	1	1	0	0	0	0	0	0	0	0	0
	Cefiderocol	5	0	0	0	0	0	0	1	1	1	1	0	1	0	0	0	0
	Cefotaxime	6	1	1	0	0	0	1	1	1	0	0	0	0	0	1	0	0
	Cefotetan	6	1	1	0	1	1	1	1	0	0	0	0	0	0	0	0	0
	Cefoxitin	6	1	1	0	1	1	1	1	0	0	0	0	0	0	0	0	0
	Ceftazidime	6	0	1	0	0	0	1	1	1	1	1	0	0	0	0	0	0
	Ceftriaxone	6	1	1	0	0	0	1	1	1	0	0	0	0	0	1	0	0
	Ceftaroline	7	1	1	0	0	0	1	1	1	0	0	0	0	1	1	0	0
	Cefepime	8	1	1	0	0	0	1	1	1	1	1	0	0	0	1	0	0
Fluoroquinolone	Ciprofloxacin	9	1	0	0	0	0	1	1	1	1	1	1	1	1	0	0	0
	Norfloxacin	9	1	0	0	0	0	1	1	1	1	1	1	1	1	0	0	0
	Ofloxacin	9	1	0	0	0	0	1	1	1	1	1	1	1	1	0	0	0
	Gemifloxacin	11	1	1	1	0	0	1	1	1	0	1	1	1	1	1	0	0
	Levofloxacin	12	1	1	1	0	0	1	1	1	1	1	1	1	1	1	0	0
	Moxifloxacin	13	1	1	1	1	1	1	1	1	0	1	1	1	1	1	0	0
	Delafloxacin	14	1	1	1	1	1	1	1	1	1	1	1	1	1	1	0	0
Folate pathway inhibitor	Sulfamethoxazole-trimethoprim	7	1	1	0	0	0	1	1	1	0	1	0	0	1	0	0	0
Fosfomycin	Fosfomycin	8	1	1	1	0	0	0	1	1	0	0	0	1	1	0	1	0
Glycopeptide	Dalbavancin	5	1	1	1	0	0	0	0	0	0	0	0	0	1	1	0	0
	Telavancin	5	1	1	1	0	0	0	0	0	0	0	0	0	1	1	0	0
	Vancomycin	5	1	1	1	0	0	0	0	0	0	0	0	0	1	1	0	0
	Oritavancin	6	1	1	1	0	0	0	0	0	0	0	0	0	1	1	1	0
Glycylcycline	Eravacycline	15	1	1	1	1	1	1	1	1	0	1	1	1	1	1	1	1
	Omadacycline	15	1	1	1	1	1	1	1	1	0	1	1	1	1	1	1	1
	Tigecycline	15	1	1	1	1	1	1	1	1	0	1	1	1	1	1	1	1
Lincosamide	Clindamycin	6	1	1	0	1	1	0	0	0	0	0	0	0	1	1	0	0
Lipopeptide	Daptomycin	6	1	1	1	0	0	0	0	0	0	0	0	0	1	1	1	0
Macrolide	Erythromycin	3	1	1	0	0	0	0	0	0	0	0	1	0	0	0	0	0
	Clarithromycin	5	1	1	0	0	0	1	0	0	0	0	1	0	1	0	0	0
	Azithromycin	6	1	1	0	1	0	1	1	0	0	0	1	0	0	0	0	0
Monobactam	Aztreonam	4	0	0	0	0	0	1	1	1	1	0	0	0	0	0	0	0
Nitrofuran	Nitrofrantoin	7	1	1	1	0	0	0	1	0	0	0	0	1	1	0	1	0
Nitroimidazole	Metronidazole	2	0	0	0	1	1	0	0	0	0	0	0	0	0	0	0	0
	Tinidazole	2	0	0	0	1	1	0	0	0	0	0	0	0	0	0	0	0
Oxazolidinone	Linezolid	6	1	1	1	0	0	0	0	0	0	0	0	0	1	1	1	0
	Tedizolid	6	1	1	1	0	0	0	0	0	0	0	0	0	1	1	1	0
Penicillin	Dicloxacillin	2	1	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0
	Nafcillin	2	1	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0
	Oxacillin	2	1	1	0	0	0	0	0	0	0	0	0	0	0	0	0	0
	Penicillin G, V	3	0	1	1	1	0	0	0	0	0	0	0	0	0	0	0	0
	Amoxicillin	5	0	1	1	1	0	1	1	0	0	0	0	0	0	0	0	0
	Ampicillin	5	0	1	1	1	0	1	1	0	0	0	0	0	0	0	0	0
	Piperacillin	7	0	1	1	1	0	1	1	1	1	0	0	0	0	0	0	0
Phenicol	Chloramphenicol	9	1	1	1	1	1	1	1	0	0	0	0	0	1	0	1	0
Pleuromutilin	Lefamulin	7	1	1	0	0	0	1	0	0	0	0	1	0	1	1	1	0
Polymyxin	Colistimethate (colistin)	6	0	0	0	0	0	1	1	0	1	1	0	1	0	0	0	1
	Polymyxin B	6	0	0	0	0	0	1	1	0	1	1	0	1	0	0	0	1
Rifampin	Rifampin	5	1	1	0	0	0	1	0	0	0	0	0	0	1	1	0	0
Streptogramin	Quinupristin-dalfopristin	5	1	1	0	0	0	0	0	0	0	0	0	0	1	1	1	0
Tetracycline	Doxycycline	7	1	1	0	0	0	1	1	0	0	0	1	0	1	1	0	0
	Tetracycline	7	1	1	0	0	0	1	1	0	0	0	1	0	1	1	0	0
	Minocycline	8	1	1	0	0	0	1	1	0	0	1	1	0	1	1	0	0

Abbreviations: A. baumannii, Acinetobacter baumannii; B. fragilis, Bacteroides fragilis; CRE, carbapenem-resistant Enterobacter spp.; E. coli, Escherichia coli; E. faecalis, Enterococcus faecalis; ESBL, extended-spectrum β-lactamase–producing Enterobacteriaceae; H. influenzae; Haemophilus influenzae; K. pneumoniae, Klebsiella pneumoniae; MRSA, methicillin-resistant S. aureus; P. aeruginosa, Pseudomonas aeruginosa; PRSP, penicillin-resistant Streptococcus pneumoniae; S. aureus, Staphylococcus aureus; VRE, vancomycin-resistant Enterococcus faecium.

ASC score is the sum of all points in the row.

Case-Based Examples for Comparisons Between DOT and DASC

Figure 1A shows an example of a switch from antibiotic combination therapy to monotherapy with a similar spectrum. When a patient was empirically treated with ceftriaxone and metronidazole (ASC scores: ceftriaxone, 6; metronidazole, 2; total, 8) and switched to ertapenem (ASC score, 9) at day 4, the daily DOT decreases from 2 to 1 despite only a slight difference in daily ASC score.

Figure 1.

Example of assessment of antibiotic use between days of therapy (DOT) and days of antibiotic spectrum coverage (DASC). A, Assessment of switching from combination therapy to monotherapy with similar spectrum. B, Assessment of de-escalation.

Figure 1B shows examples of de-escalation. When a patient was empirically treated for 7 days with a combination of vancomycin and piperacillin-tazobactam (ASC scores: vancomycin, 5; piperacillin-tazobactam, 11; total, 16), the total of DOT and DASC are 14 and 112, respectively. If the antibiotic therapy was subsequently de-escalated from vancomycin and piperacillin-tazobactam to ceftriaxone and metronidazole (total ASC score, 8) on day 4, the total DOT remains 14, but the total DASC changes from 112 to 80, reflecting the de-escalation effects.

Study Population for Comparison Between DOT and DASC for Hospital Benchmarking

Characteristics of included hospitals are summarized in Table 2. There were 616 357 hospital admissions (371 507 unique patients) at 124 VA hospitals during the study period, reflecting a total of 3 888 879 DP. Among those unique patients, 187 585 (50.5%) received ≥1 day of antibiotic therapy during their inpatient stay. The median and interquartile range of unique number of patients, percentage of ICU cases, percentage of surgical patients, DOT, DASC, DASC per DOT, DOT per 1000 DP, and DASC per 1000 DP at each hospital are shown in Table 2.

Table 2.

Characteristics of Hospitals Included for Benchmarking and Metrics for Antibiotic Use

Characteristic or Metric	Median (IQR)a
Hospital characteristic
No. of operating beds	74 (34–107)
Teaching hospitals, no. (%)	94 (75.8)
Rural hospitals, no. (%)	15 (12.1)
Transplant programs, no. (%)	11 (8.9)
Care volume at each hospital
No. of unique patients	2941 (1384–4393)
DP	27 400 (12 204–42 941)
Proportion in ICU, %	15.1 (11.7–19.6)
Proportion in surgical units, %	19.2 (12.5–27.6)
Antibiotic use metric
DOT	11 455 (5482–19 482)
DASC	78 083 (40 015–135 512)
DASC/DOT	7.02 (6.70–7.23)
DOT/1000 DP	440.7 (387.2–496.1)
DASC/1000 DP	3107.4 (2578.3–3477.9)

Abbreviations: DASC, days of antibiotic spectrum coverage; DOT, days of therapy; DP, days present; ICU, intensive care unit; IQR, interquartile range.

Data represent median (IQR) unless otherwise specified.

Relationship Between DOT per 1000 DP and DASC per DOT

The relationships between DOT per 1000 DP and DASC per DOT are shown in Figure 2. There was little correlation between DOT per 1000 DP and DASC per DOT (Spearman rank correlation test, ρ = 0.16).

Figure 2.

Distribution of days of therapy (DOT) per 1000 days present (DP) and days of antibiotic spectrum coverage (DASC) per DOT. Vertical line represents median value of DOT per 1000 DP; horizontal line, median value of DASC per DOT.

Comparison of Benchmarking Based on DASC and DOT

Figure 3 shows the distribution of DOT and DASC per 1000 DP based on the ranking of DOT per 1000 DP. The range of ranking difference percentages was from −29% (DASC ranking was higher than DOT ranking) to 25% (DOT ranking was higher than DASC ranking) (Supplementary Figure 1). Twenty-seven hospitals (21.8%) had a difference of ≥10% between their rankings on the 2 metrics. Stratified analyses of ICU versus non-ICU and medical versus surgical units showed more ranking changes in ICU and surgical units, suggesting that variabilities in the average spectrums of antibiotic therapies used in those units might be larger (Supplementary Figure 2).

Figure 3.

Comparison of benchmarking between days of antibiotic spectrum coverage (DASC) and days of therapy (DOT) per 1000 days present (DP).

DISCUSSION

We developed a novel metric for antibiotic consumption that uses antibiotic spectrum scores and compared it with a DOT-based antibiotic use assessment. Compared with DOT-based benchmarking, 21.8% of hospitals had >10% differences in their ranking after DASC was applied. The lack of correlation between DASC per DOT and DOT per 1000 DP suggests that the efforts of ASPs to use narrower-spectrum agents through judicious empiric therapy or de-escalation are not reflected in DOT-based metrics. In addition, DASC is robust to the difference between monotherapy and combination therapy regimens with a similar spectrum, which has the benefit of avoiding inadvertently penalizing hospitals that use appropriate combination therapies. This novel benchmark allows more comprehensive assessments of antibiotic use across hospitals than DOT as a composite measure of 2 aspects, the total volume of antibiotic consumption and the average spectrum of coverage, and can provide hospital benchmarking for inpatient antibiotic consumption with more construct validity [15].

To inform more meaningful ASP evaluations, antibiotic use metrics should reflect the efforts of each hospital’s ASP. Currently, DOT and SAAR are commonly used metrics used for benchmarking, but neither of those metrics directly reflect efforts to narrow the spectrum of the antibiotic being prescribed. NHSN also provides SAAR for subcategories of antimicrobial agents (eg, broad-spectrum agents predominantly used for community-acquired infections or antibacterial agents posing the highest risk for Clostridioides difficile infections), but they do not directly provide intuitively understandable indicators for de-escalation or other spectrum-narrowing efforts [2]. In addition, those categories are not mutually exclusive, and simultaneous interpretation of SAARs from different categories can be challenging.

Although improvements to address overlaps among antimicrobial categories have been proposed [16], they are yet to be incorporated into the NSHN AUR module. Hence, hospitals that devote more efforts to use narrower-spectrum therapy through judicious empiric therapy or de-escalation may not receive easily understandable and fair evaluations based on those conventional metrics. Unfair assessments may lead to a false understanding of the quality of antibiotic use in each hospital. This issue is particularly important when considering public reporting, which has already occurred with other healthcare quality metrics [17]. DASC could potentially improve the reliability of benchmarking and have positive effects on program assessments of ASPs.

There have been several reports on scoring the spectrum of antibiotics [9, 18–21]. Among them, reports from Madaras-Kelly et al [19] and Gerber et al [9] have set up comprehensive scoring systems. The antibiotic spectrum score, as reported by Madaras-Kelly et al [19] included 27 antibiotics and 14 spectrums. However, the classification of antibiotic agents was limited to the most frequently used antibiotic agents in relation to the disease (eg, healthcare-associated pneumonia), and the list of antibiotic agents was not exhaustive. Weighted scores for each organism-agent combination were assigned based on the likelihood of antibiotic coverage from epidemiologic data within VHA hospitals.

This approach provides objectivity and reflects antibiotic efficacy in clinical settings but also raises a potential need for continuous updates due to temporal changes in the epidemiology of antibiotic resistance. In addition, the generalizability to other geographic regions with different prevalence of antibiotic resistance can be questionable. Gerber et al [9] proposed a simple yet practical antibiotic spectrum scoring method, ASI, which assigns a dichotomous score to each antibiotic-organism combination. However, the rationale for scoring decisions was not provided in their report, and the epidemiology of antibiotic resistance can potentially warrant revisions of scores over time. The ASI list included 49 commonly used antibiotic agents for 14 antibiotic spectrums but was not comprehensive enough to be operationalized for hospital benchmarking.

Another important advantage of our novel benchmarking metric, DASC, is that the spectrum evaluations were clearly defined into 2 categories: intrinsic resistance and acquired resistance with specific mechanisms. We assigned the dichotomous score according to the activity against naive organisms and specific resistance mechanisms, based on systematic literature searches and reviews by the panel of experts, which included infectious diseases specialists, microbiologists, and infectious diseases pharmacists. Acquired resistance was assessed based on the specific resistance mechanisms for each agent rather than the epidemiology of those organisms. Therefore, it is less likely that the ASC score needs to be updated to adapt to external factors, such as the change in the prevalence of antibiotic resistance unless new antibiotic agents become available for clinical uses. Because DASC can account for both quantities and spectrums of antibiotic therapy and reflects more scopes of ASP, it can replace DOT as a foundational measurement of antibiotic consumption and serve as a basis for future risk-adjusted metrics.

The relationship between DOT per 1000 DP and DASC per DOT showed that low DOT per 1000 DP does not necessarily mean high narrow-spectrum antibiotic use. Although it is difficult to assess without risk adjustments by case mix and severity of illness, hospitals with a high DASC per DOT may need to look for opportunities to avoid unnecessary use of broad-spectrum agents, while hospitals with a low DASC per DOT may need to review appropriate treatment duration or evaluate the frequency with which combination therapy is prescribed. Adding spectrum information to the evaluation of antibiotic use can provide an additional dimension in evaluations of ASPs and facilitate more detailed interhospital comparisons of antibiotic use.

Our study has a few limitations. First, we did not perform a qualitative evaluation (ie, evaluating appropriateness of therapy) other than the spectrum of antibiotics, as this would have required manual case-based reviews to determine the appropriateness of therapy for individual cases. Second, DASC still inherits 2 limitations of DOT, the evaluation of extended interval administrations for renal or hepatic dysfunction and the duplicated counts for same-day administrations of multiple regimens when a therapeutic switch occurs [2]. Third, our study did not consider patient-level case-mix or prevalence of antibiotic resistance at each hospital for fair benchmarking, except for stratified analyses. Future studies should assess the impact of case mix or antimicrobial resistance prevalence on DASC and develop DASC-based risk-adjusted metrics for benchmarking with higher content validity. Finally, we assumed that antibiotics of the same spectrum are equal in the selective pressure they exert without providing weights in the scoring. Some broad-spectrum antibiotics might lead to more emergence or spread of antibiotic resistance than others, although the current knowledge concerning comparative effects on selection pressures is sparse.

In conclusion, we propose a novel quantitative benchmarking metric for inpatient antibiotic consumption, DASC, with additional consideration of the antibiotic spectrum. Compared with DOT, DASC can simultaneously evaluate ASP efforts to select agents with a narrower therapeutic spectrum and to reduce overall consumption, while not inappropriately penalizing hospitals that use combination therapeutic regimens ,which have a spectrum similar to that of monotherapy regimens. DASC changed benchmarking rankings by >10% at approximately 20% of hospitals, suggesting it provides an added dimension that may help enable more accurate hospital comparisons. Further studies are needed to evaluate whether hospital benchmarking with the DASC metric can lead to improved antimicrobial stewardship and reduced selection of antibiotic-resistant pathogens.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Disclaimer. The views expressed herein are those of the authors and do not necessarily reflect the views of the US Department of Veterans Affairs or the Agency for Healthcare Research and Quality. The funding sources had no role in the design and conduct of the study; collection, management, and analysis of the data; or preparation, review, and approval of the manuscript.

Financial support. This work was supported by the Veterans Health Administration (VHA) National Center for Patient Safety through the Patient Safety Center for Inquiry program, the Agency for Healthcare Research and Quality (grant K08HS027472 to M. G.), and the VHA Health Services Research and Development Service through the Center for Access and Delivery Research and Evaluation (CADRE) (grant CIN 13-412 to E. N. P.).

Author notes

Potential conflicts of interest. E. N. P. reports support from the US Department of Veterans Affairs (grants QUE 15-269 and RVR 19-477), outside the conduct of the study. D. J. D. reports grants to their institution from bioMerieux, to support clinical trials, and individual payments from OpGen, for consulting on novel diagnostics, and JMI Laboratories, for consulting on antibiotic resistance surveillance studies. D. I. reports payments to their institution from Gilead Sciences, for clinical trials on remdesivir, and from Leidos, for Adaptive COVID-19 Treatment Trial 3 and Adaptive COVID-19 Treatment Trial 4, clinical trials sponsored by the National Institutes of Health; D. I. also serves on the data and safety monitoring board for Evergreen Pharma. P. K. reports grants from Accelerate Diagnostics to University of Iowa Hospitals & Clinics principal investigators to support clinical study, outside the submitted work, and reports serving on an advisory panel for Gilead Sciences, on remdesivir in coronavirus disease 2019. M. G. reports support for the present study from the VHA Patient Safety Center for Inquiry program. D. J. L. reports support from the Health Services Research and Development Service (grant 20-240 [coinvestigator], career development award 16-204, and grant D 20-280 [principal investigator: D. J. L.]), the Centers for Diseases Control and Prevention EpiCenter (grant J207400-G; coinvestigator), and the Merck Investigator Studies Program. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.