Antimicrobials are one of the most commonly prescribed medications in veterinary medicine,1 and although their use has resulted in substantial improvements in animal health and welfare, such use can also result in antimicrobial resistance among surviving microorganisms, resultant treatment failures, increased morbidity and mortality rates,2,3 and potential transmission of resistant microorganisms to people.4–6 Unlike in human and small animal medicine, antimicrobial prescribing guidelines are generally not available for clinically important infectious diseases in large animals, although broad, general guidelines for antimicrobial use in veterinary medicine exist.7 Many studies,8–15 including those conducted at our institution, have documented use of antimicrobials in veterinary hospitals inconsistent with general guidelines for appropriate antimicrobial prescribing practices. Nevertheless, surveys have shown that veterinarians are growing increasingly concerned about antimicrobial resistance and interested in promoting the judicious use of antimicrobials.12,13,16–18
A fundamental requirement for improving antimicrobial prescribing practices within an institution is understanding how antimicrobials are currently being used. When antimicrobial use guidelines for specific conditions exist, then inappropriate use can be identified. For example, appropriateness of antimicrobial use has been assessed in dogs and cats with urinary tract infections or respiratory disease,13,19 for which antimicrobial use guidelines are available.20,21 Other studies15,22–26 have examined whether antimicrobial prescribing by clinicians adheres to broad recommendations for judicious antimicrobial use, such as those proposed by the American College of Veterinary Internal Medicine7 or the AVMA.27 When disease-specific antimicrobial use guidelines are unavailable, as is the situation in large animal medicine, differences in prescribing patterns among clinicians could indicate a lack of consensus regarding appropriate antimicrobial use and signal where antimicrobial use guidelines are needed.
Data on antimicrobial use by individual clinicians can also be used to provide feedback to clinicians on their prescribing practices, which, in human medicine, has been shown to improve patient outcomes, shorten the duration of hospitalization, decrease the rate of antimicrobial prescribing deemed unnecessary, and save money in the inpatient setting.28,29 Several studies12,22,30–35 have elucidated the drivers of antimicrobial prescribing by veterinary clinicians as a group, and others22,24,25,36,37 have provided some information on antimicrobial prescribing by individual clinicians, but none have provided quantitative assessments of antimicrobial prescribing or assessed factors associated with prescribing.
The New Bolton Center hospital at the University of Pennsylvania School of Veterinary Medicine is a teaching hospital for large animals, with 7 species (equine, bovine, caprine, ovine, porcine, camelid, and cervid) and > 200 breeds represented in the patient population. Understanding antimicrobial prescribing practices in this type of institution would be particularly important because the prescribing behaviors of veterinary clinicians at a teaching hospital influence the prescribing behavior of future veterinary clinicians. The purpose of the study reported here was to characterize antimicrobial prescribing practices at the level of the clinical unit and individual clinician in this hospital by use of administrative and billing databases and to determine factors associated with the prescribing of antimicrobials in general, the prescribing of critically important antimicrobials specifically, and the quantity of antimicrobials prescribed per visit. We believe that such information would be important as a first step in the development of a quality improvement initiative involving clinician feedback and interventions designed to improve antimicrobial prescribing practices through changing behavior.
Materials and Methods
Data collection
A retrospective cross-sectional study was conducted involving all patients evaluated at the New Bolton Center hospital of the University of Pennsylvania School of Veterinary Medicine from January 1, 2013, to August 1, 2018, and associated electronic administrative and billing records. Animals that were dead on arrival and accompanying healthy (companion) animals were excluded, and all visits were treated as independent events (ie, some animals could have been represented more than once). Information was extracted from the records regarding patient signalment, visit date, hospital service and clinician at hospital admission and discharge, initial complaint (free-text data), diagnoses (coded data), and procedures performed during the visit, including antimicrobials dispensed. To truly reflect antimicrobial prescribing practices at the individual clinician level, the clinician was defined as the person who ordered the medication rather than the person who admitted or discharged the patient and the service was defined as the service to which this clinician belonged.
All procedure codes used in the records system (n = 5,023) were manually reviewed by a veterinary clinician (LER) to identify antimicrobial procedure codes (228), and those codes were categorized by antimicrobial class (Appendix). The CIAHPs, as defined by the World Health Organization38 (ie, third-generation cephalosporins, glycopeptides, macro-lides, ketolides, polymyxins, and quinolones), were also identified. All possible entries for the “diagnosis” and “presenting complaint” fields were reviewed and categorized by affected body system.
Statistical analysis
Patient characteristics of all animals visiting the hospital during the study period were compared across species and hospital service by means of ANOVA (normally distributed data) or the Kruskal-Wallis test (nonnormally distributed data). For visits that lacked a body weight recording in the administrative database, an estimated weight was assigned for adult (≥ 1-year-old) patients on the basis of the mean weight of other patients of the same species in our database. When data for calculation of mean body weights were unavailable, missing weights were assigned a value on the basis of the patient's species, age, and breed by use of mean body weights reported in the literature.39–41 For patients < 1 year old, body weight was calculated as a function of species and age by use of published species-specific growth equations.42–44
For analyses involving antimicrobial prescribing practices, only visits involving the 3 hospital services with the highest caseloads and clinicians belonging to 1 of these 3 services who regularly attended to patients (ie, ≥ 50 patients/y) were included. Three metrics were used to describe prescribing patterns: the proportion of visits in which at least 1 antimicrobial or CIAHP was dispensed (ie, frequency of use), the number of ADDs per visit (ie, duration of use), and the number of antimicrobials (ie, number of distinct classes of antimicrobials or types of formulations) dispensed per patient (ie, distribution of use). For example, if penicillin and gentamicin (2 distinct antimicrobial classes) were dispensed for a given patient, the number of antimicrobials dispensed would be considered 2. If an injectable metronidazole formulation was dispensed for a given patient during hospitalization and then an oral metronidazole formulation was dispensed for administration after hospital discharge, the number of antimicrobials dispensed would also be considered 2. However, because there were few situations when different formulations of the same antimicrobial were dispensed for the same patient, the number of antimicrobials generally represented the number of distinct antimicrobial classes dispensed. The proportion of visits in which an antimicrobial was dispensed and the number of antimicrobials were determined for all formulations of antimicrobial (ie, injectable, oral, topical, ophthalmic, and intramammary). The number of ADDs was calculated only for injectable and oral antimicrobial formulations as follows:
where the units of active ingredient was calculated as the product of the number of units dispensed (eg, total volume [liquids] or number of tablets or capsules) and the concentration of the product (eg, mg/mL, U/mL, or mg/tablet), and the DDDvet represents the daily dose (mg/kg or U/kg) an animal should receive in 1 day as indicated by the drug label or convention at our hospital. For example, for horses, penicillin would be prescribed for administration at a dosage of 22,000 U/kg, every 6 hours, and gentamicin would be dispensed for administration at a dosage of 8.8 mg/kg, every 24 hours. The DDDvet for these 2 antimicrobials would consequently be 88,000 U/kg and 8.8 mg/kg, respectively. Therefore, if a total of 440,000,000 U of penicillin was dispensed for a 500-kg horse, the number of ADDs dispensed would be 440,000,000 U/(500 kg × 88,000 U/kg/d) −10, representing approximating 10 days of treatment with this medication. The DDDvet values used in the study were summarized (Supplementary Appendix S1, available at avmajournals.avma.org/doi/suppl/10.2460/ajvr.81.2.103).
Values of these 3 metrics were then compared across the 3 hospital services and across included clinicians within a service for all patients and for patients with the most commonly affected body systems (gastrointestinal or musculoskeletal). Proportions were compared by use of the χ2 test, and the number of ADDs per patient was compared via linear regression, with the individual clinician treated as a fixed effect. Pairwise comparisons were performed for all clinicians within a service by use of marginal linear predictions and the Sidak correction for multiple comparisons.
To examine potential associations of independent variables related to the patient and visit (patient species, age, system affected, duration of hospitalization, and whether bacterial culture had been performed) with dispensing of an antimicrobial or CIAHP, number of antimicrobials dispensed, and number of ADDs dispensed per patient (ie, the outcome or dependent variables), multilevel mixed-effects logistic regression (yielding ORs), negative binomial regression (yielding IRRs), and linear regression (yielding coefficients), respectively, were performed. Patient and visit factors were treated as fixed effects, whereas clinician and service (but not patient) were treated as random effects. Variables considered a priori to be important or potential confounders were added to the models in a stepwise fashion. Variables that substantially altered the association between the evaluated outcome variable and previously added independent variables (ie, variables that resulted in a change of 20% or more in the OR for 1 or more independent variables) were considered confounders and were retained. Fits of nested models (ie, different versions with the same outcome variable) were compared via examination of the Akaike and Bayesian information criterion values. Models with the lowest of these values were retained as final models.
All data analyses were conducted with the aid of statistical software.a Values of P < 0.05 were considered significant.
Results
Hospital visits and patients
From 2013 through 2018, a total of 16,625 animals were evaluated at 23,428 visits (Table 1). A total of 106 clinicians across 12 services were represented in our database. Of the 106 clinicians, 24 (23%) who regularly attended to patients (ie, ≥ 50 patients/y) were considered in the statistical analyses regarding antimicrobial prescribing practices. The Emergency and Critical Care (n = 2,995 animals and 3,161 visits), Medicine and Ophthalmology (5,555 animals and 6,743 visits), and Surgery (7,459 animals and 10,083 visits) services had the highest caseloads and were therefore used in these analyses. Overall, no body weight was recorded for 7,894 (33.7%) visits, and therefore body weight estimates were used instead. A diagnosis or owner complaint was missing for 2,919 of 23,428 (12.5%) visits.
Number (%) of all animals evaluated by various services at the New Bolton Center hospital of the University of Pennsylvania from 2013 through 2018 and median (IQR) patient age and duration of hospitalization.
| Service | ||||||
|---|---|---|---|---|---|---|
| Species | Emergency and Critical Care | Medicine and Ophthalmology | Surgery | Whole hospital* | Age (y) | Duration of hospitalization (d) |
| Equine | 2,403 (76.0) | 5,432 (80.6) | 8,261 (81.9) | 19,366 (82.7) | 8.6 (4.0–14.0) | 1 (0.5–4) |
| Bovine | 238 (7.5) | 341 (5.1) | 776 (7.7) | 1,398 (6.0) | 1.6 (0.2–4.0) | 2 (0.5–5) |
| Caprine | 315 (10.0) | 550 (8.2) | 481 (4.8) | 1,379 (5.9) | 2.0 (0.4–5.0) | 3 (0.5–8) |
| Ovine | 105 (3.3) | 176 (2.6) | 166 (1.6) | 519 (2.2) | 1.0 (0.2–4.5) | 2 (0.5–6) |
| Porcine | 46 (1.5) | 116 (1.7) | 328 (3.3) | 505 (2.2) | 1.6 (0.4–4.2) | 1 (0.5–4) |
| Camelid | 51 (1.6) | 126 (1.9) | 69 (0.7) | 253 (1.1) | 3.4 (0.5–7.9) | 3 (0.5–7) |
| Cervid | 4 (0.1) | 2 (0.03) | 2 (0.02) | 8 (0.03) | 0.7 (0.3–2.4) | 1.3 (0.5–6) |
| Total | 3,162 (100) | 6,743 (100) | 10,083 (100) | 23,428 (100) | 7.0 (3.0–12.9) | 2 (0.5–4) |
Data in this column represent the 3 hospital services with the highest caseloads (Emergency and Critical Care, Medicine and Ophthalmology, and Surgery) plus other services that accounted for < 15% of visits.
Equids, bovids, and goats were the most common species evaluated by all 3 hospital services and overall (Table 1). For almost all categories of affected body systems, the proportion of visits involving a particular body system differed significantly (P < 0.001) by hospital service (Table 2). Within a service, significant (P < 0.001) heterogeneity was evident in the distribution of patient age and in the proportions of visits associated with the different affected body systems (ie, types of clinical signs) by clinician. Submission of specimens for anaerobic or aerobic bacterial culture was documented for 3,096 (13.2%) visits, and positive results of culture and subsequent antimicrobial susceptibility testing were obtained for 1,808 visits (7.7%).
Number (%) of visits by the animals of Figure 1, by affected body system and hospital service.
| Body system | Whole hospital* | Emergency and Critical Care | Medicine and Ophthalmology | Surgery |
|---|---|---|---|---|
| Oropharyngeal or nasal | 1,767 (7.5) | 71 (2.3) | 204 (3.0) | 1,334 (13.2) |
| Ocular | 2,602 (11.1) | 101 (3.2) | 2,363 (35.0) | 119 (1.2) |
| Respiratory | 833 (3.5) | 154 (4.9) | 462 (6.9) | 181 (1.8) |
| Cardiovascular | 696 (3.0) | 120 (3.8) | 252 (3.7) | 70 (0.7) |
| Gastrointestinal | 3,116 (13.3) | 1,603 (50.7) | 866 (12.8) | 609 (6.0) |
| Urogenital | 3,308 (14.1) | 302 (9.6)† | 664 (9.8) | 983 (9.7)† |
| Musculoskeletal | 5,542 (23.7) | 161 (5.1) | 469 (7.0) | 4,498 (44.6) |
| Integumentary | 1,185 (5.0) | 275 (8.7) | 183 (2.7) | 705 (7.0) |
| Neurologic | 325 (1.4) | 38 (1.2)† | 196 (2.9)† | 83 (0.8) |
| Other or unclassifiable | 2,866 (12.2) | 228 (7.2) | 772 (11.5)† | 1,091 (10.8)† |
| Healthy | 1,188 (5.1) | 108 (3.2)† | 312 (4.6) | 411 (4.1)† |
| Total | 23,428 (100) | 3,162 (100) | 6,744 (100) | 10,084 (100) |
Within a row, the indicated percentages do not differ significantly (P > 0.05; χ2 test) from each other. See Table 1 for remainder of key.
Proportions of visits in which an antimicrobial was dispensed
A total of 44,410 instances of antimicrobials being dispensed were recorded, and at least 1 antimicrobial was dispensed in 9,853 of 23,428 (42.1%) visits. Of the 3 evaluated services, the Emergency and Critical Care service was responsible for the highest proportion of dispensings and the Surgery service had the lowest (Table 3; P = 0.001). Significant (P < 0.001) heterogeneity was identified when comparing clinicians within a service with respect to data for all patients, patients with disease affecting the gastrointestinal system, and patients with disease affecting the musculoskeletal system, although pairwise comparisons revealed statistically similar proportions among certain pairs of clinicians (Figure 1). By clinician, proportions ranged from 26% to 62% for the Surgery service, from 28% to 62% for the Medicine and Ophthalmology service, and from 45% to 62% for the Emergency and Critical Care service. For individual clinicians, these proportions were fairly consistent over time, with the highest between-year difference being 6 percentage points. Among clinicians belonging to the Surgery service, there was greater consistency in the proportions of visits in which antimicrobials were dispensed for patients with disease affecting the gastrointestinal system than for all patients. This association was not observed for the other 2 services. In contrast, heterogeneity was again evident among clinicians within a service for patients with disease affecting the musculoskeletal system.
Summary of antimicrobial prescribing practices by hospital service for the animals of Table 1.
| Service | Proportion (%) of visits in which an antimicrobial was dispensed* | Proportion (%) of visits with an antimicrobial dispensed in which a CIAHP was dispensed* | Median (IQR) No. of ADDs per visit | Mean (SD) number of antimicrobials dispensed per visit |
|---|---|---|---|---|
| Emergency and Critical Care | 1,962/3,161 (62.1) | 434/1,962 (22.1) | 6.3 (2.4–16.2) | 2.5 (1.4) |
| Medicine and Ophthalmology | 3,080/6,743 (45.7) | 975/3,080 (31.7) | 2.3 (0–10.7) | 1.8 (1.2) |
| Surgery | 4,394/10,083 (43.6) | 735/4,394 (16.7) | 3.3 (1.0–9.8) | 2.0 (1.2) |
| Whole hospital† | 9,853/23,428 (42.1) | 2,360/9,853 (24.0) | 3.6 (0.8–11.1) | 2.0 (1.3) |
These proportions differed significantly (P < 0.00I) across the 3 hospital services.
Data in this row represent the 3 hospital services with the highest caseloads (Emergency and Critical Care, Medicine and Ophthalmology, and Surgery) plus other services that accounted for < I5% of visits.
Proportion of visits in which any antimicrobial was dispensed (height of the dark gray portion; n = 9,436 visits) and any CIAHP was dispensed (height of the light gray portion; 2,144) by individual clinicians, grouped by the hospital service to which the clinicians belonged, for animals (bovids, equids, sheep, goats, swine, camelids, and cervids) evaluated at the New Bolton Center hospital of the University of Pennsylvania from 20l3 through 2018. Panels show data for all patients (A), only patients with gastrointestinal disease (B), and only patients with musculoskeletal disease (C).a–c Within a service, values for clinicians with the same letters do not differ significantly (P > 0.05).
Citation: American Journal of Veterinary Research 81, 2; 10.2460/ajvr.81.2.103
Patient species, age, and affected body system as well as submission of specimens for bacterial culture were all significantly associated with the likelihood of receiving an antimicrobial (Table 4). Bovids and small ruminants were significantly (P < 0.001) more likely to have an antimicrobial dispensed than were equids (OR, 2.35 and 1.37, respectively). Submission of specimens for bacterial culture was associated with an increased odds of having an antimicrobial dispensed (OR, 2.47). Animals with disease affecting the ocular or integumentary system were most likely to have an antimicrobial dispensed, compared with healthy animals (OR, 16.10 and 14.00, respectively), although animals with any body system affected were more likely than healthy animals to have an antimicrobial dispensed (P < 0.001).
Associations of patient factors with the likelihood of having an antimicrobial or CIAHP dispensed at a visit* and the number of distinct antimicrobial classes or formulations dispensed† and number of ADDs (injectable and oral formulations only)‡ per visit for the animals of Table 1.
| Likelihood of having an antimicrobial dispensed (n = 23,428 visits) | Likelihood of having a CIAHP dispensed (n = 23,428 visits) | No. of antimicrobials dispensed (n = 9,853 visits) | No. of ADDs dispensed (n = 9,853 visits) | |||||
|---|---|---|---|---|---|---|---|---|
| Factor | OR (95% CI) | P value | OR (95% CI) | P value | IRR (95% CI) | P value | Coefficient (95% CI) | P value |
| Species | ||||||||
| Equine | Referent | — | Referent | — | Referent | — | Referent | — |
| Bovine | 2.35 (1.97–2.79) | < 0.001 | 1.85 (1.48–2.31) | < 0.001 | 1.16 (1.14–1.37) | < 0.001 | 0.65 (−0.34 to 1.64) | 0.20 |
| Small ruminant | 1.37 (0.97–1.46) | < 0.001 | 1.41 (1.14–1.73) | 0.001 | 0.86 (0.76–0.90) | < 0.001 | 8.76 (7.92–9.60) | < 0.001 |
| Other | 1.19 (1.18–1.58) | 0.08 | 2.36 (1.77–3.14) | < 0.001 | 0.97 (0.84–1.10) | 0.38 | 8.18 (7.25–9.74) | < 0.001 |
| Age (y) | 0.93 (0.96–0.98) | < 0.001 | 0.97 (0.96–0.98) | < 0.001 | 0.98 (0.97–0.99) | < 0.001 | −0.05 (−0.09 to −0.02) | 0.002 |
| Duration of hospitalization (wk) | — | — | 1.11 (1.07–1.15) | < 0.001 | 1.30 (1.28–1.33) | < 0.001 | 2.99 (2.85–3.14) | < 0.001 |
| Specimens submitted for bacterial culture (yes vs no) | 2.47 (2.23–2.73) | < 0.001 | 1.71 (1.49–1.97) | < 0.001 | 1.72 (2.00–2.24) | < 0.001 | 5.61 (4.99–6.24) | < 0.001 |
| Affected body system | ||||||||
| Healthy | Referent | — | Referent | — | Referent | — | Referent | — |
| Oropharyngeal or nasal | 9.93 (7.58–13.00) | < 0.001 | 3.93 (1.63–7.45) | 0.002 | 7.53 (6.05–9.35) | < 0.001 | 7.23 (5.90–8.56) | < 0.001 |
| Ocular | 16.10 (11.90–21.90) | < 0.001 | 5.50 (2.27–13.40) | < 0.001 | 8.77 (6.97–11.00) | < 0.001 | 5.36 (3.99–6.72) | < 0.001 |
| Respiratory | 10.30 (7.69–13.70) | < 0.001 | 7.81 (3.27–18.70) | < 0.001 | 8.30 (6.66–10.40) | < 0.001 | 11.40 (9.98–12.90) | < 0.001 |
| Cardiovascular | 10.90 (7.88–15.00) | < 0.001 | 3.93 (1.59–9.67) | 0.003 | 7.64 (6.04–9.65) | < 0.001 | 8.03 (6.31–9.97) | < 0.001 |
| Gastrointestinal | 7.32 (5.68–9.43) | < 0.001 | 5.27 (2.24–12.40) | < 0.001 | 6.88 (5.59–8.47) | < 0.001 | 6.99 (5.81–8.17) | < 0.001 |
| Urogenital | 11.40 (8.82–14.70) | < 0.001 | 2.11 (0.89–5.04) | 0.09 | 6.87 (5.59–8.48) | < 0.001 | 6.47 (5.25–7.69) | < 0.001 |
| Musculoskeletal | 4.54 (3.53–5.85) | < 0.001 | 2.61 (1.09–6.24) | 0.03 | 4.83 (3.92–5.96) | < 0.001 | 4.36 (3.21–5.51) | < 0.001 |
| Integumentary | 14.00 (10.60–18.40) | < 0.001 | 1.40 (0.59–3.47) | 0.44 | 9.44 (7.62–11.69) | < 0.001 | 8.44 (7.09–9.79) | < 0.001 |
| Neurologic | 6.00 (4.22–8.50) | < 0.001 | 2.08 (0.77–5.61) | 0.15 | 5.22 (4.02–6.77) | < 0.001 | 5.48 (3.56–7.40) | < 0.001 |
| Other | 4.42 (3.43–5.71) | < 0.001 | 2.93 (1.23–7.00) | 0.02 | 4.49 (3.63–5.56) | < 0.001 | 4.39 (3.18–5.60) | < 0.001 |
Evaluated via mixed-effects multivariable logistic regression.
Evaluated via mixed-effects multivariable binomial regression.
Evaluated via mixed-effects multivariable linear regression.
— = Not applicable.
Proportions of visits in which a CIAHP was dispensed
Of the 9,853 visits in which an antimicrobial was dispensed (regardless of hospital service), a CIAHP was dispensed in 2,360 (24.0%) visits. This proportion differed significantly (P < 0.001) by service and was highest for the Medicine and Ophthalmology service (31.7% [975/3,080]) and lowest for the Surgery service (16.7% [734/4,394]). For individual clinicians, the proportion of visits in which a CIAHP was dispensed ranged from 3.3% to 100% (data not shown), and similar variability was also observed for patients with disease affecting the gastrointestinal or musculoskeletal system (Figure 1). Each nonequid species (bovids, small ruminants, camelids, swine, and cervids) was significantly (P < 0.001) more likely than equids to have a CIAHP dispensed (Table 4). Submission of specimens for bacterial culture was associated with a 71% greater odds of having a CIAHP dispensed (P < 0.001). Each 1-year increase in age was associated with a 3% decrease in the odds of having a CIAHP dispensed (P < 0.001), whereas each 1-week increase in duration of hospitalization was associated with an 11% increase in these odds (P < 0.001). Animals with disease affecting the respiratory system were most likely to receive a CIAHP, compared with healthy animals (OR, 7.81; P < 0.001).
Number of ADDs per patient
The median (IQR) number of ADDs per patient for the whole hospital during the study period was 3.6 (0.8 to 11.1; Table 3). This number was highest within the Emergency and Critical Care service and lowest within the Medicine and Ophthalmology service and differed significantly (P < 0.001) among clinicians within services for all patients and for patients with disease affecting the gastrointestinal or musculoskeletal system, although some consistency in these numbers was observed for certain pairs of clinicians within services (Figure 2).
Box plots showing the number of ADDs of antimicrobial per visit for the animals of Figure 1, by hospital service (Surgery [dark gray bars], Emergency and Critical Care [medium gray bars], and Medicine and Ophthalmology [light gray bars]) and clinician. The central horizontal bar within each box represents the median value, and the box itself represents the IQR. See Figure 1 for remainder of key.
Citation: American Journal of Veterinary Research 81, 2; 10.2460/ajvr.81.2.103
Patient species, age, affected body system, duration of hospitalization, and submission of specimens for bacterial culture were significantly associated with the number of ADDs per patient (Table 4). Small ruminants and other species had 8.76 and 8.18 as many ADDs dispensed, respectively, as did equids (P < 0.001). For every 1-year increase in age, the number of ADDs per patient decreased by 0.05 (P = 0.002), and for every 1-week increase in duration of hospitalization, this number increased by 2.99 (P < 0.001). Submission (vs no submission) of specimens for bacterial culture was associated with a 5.61-units increase in the number of ADDs per patient (P < 0.001). Animals with disease affecting the respiratory system had the greatest number of ADDs, compared with healthy animals (OR, 11.40; P < 0.001).
Number of antimicrobials dispensed
A mean (SD) of 2.0 (1.3) distinct classes or formulations of antimicrobials was dispensed per patient within the whole hospital during the study period (Table 3). This number was lowest within the Medicine and Ophthalmology service and highest within the Emergency and Critical Care service. It differed significantly (P < 0.001) across all clinicians, ranging from 1.5 to 2.7, although there was more consistency in values among clinicians within the same service than there was for other metrics (Figure 3). The number of antimicrobials dispensed was significantly associated with patient species, age, affected body system, and duration of hospitalization as well as submission of specimens for bacterial culture (Table 4). Bovids had significantly (P < 0.001) higher and small ruminants significantly (P < 0.001) lower numbers dispensed, compared with equids. For every 1-week increase in duration of hospitalization, the number of antimicrobials dispensed increased by 30%, and for every 1-year increase in patient age, it decreased by 2%. Submission (vs no submission) of specimens for bacterial culture was associated with a significant (P < 0.001) increase in the number of antimicrobials dispensed. Animals with disease affecting the integumentary system had the greatest increase in the number of antimicrobials dispensed relative to that for healthy animals (IRR, 8.44; P < 0.001), although animals with any body system affected had a greater number of antimicrobials dispensed than did healthy animals (P < 0.001).
Box-and-whisker plots of the number of antimicrobials (ie, distinct classes of antimicrobials or types of formulations) dispensed per visit for the animals of Figure l, by hospital service and clinician. The horizontal bar (within or outside boxes) represents the median value, the box represents the IQR. The ends of whiskers represent upper and lower adjacent values (ie, all data points within 1.5 times the IQR), and dots represent outlier values. See Figure 2 for remainder of key.
Citation: American Journal of Veterinary Research 81, 2; 10.2460/ajvr.81.2.103
Two or more classes of antimicrobials were dispensed in 5,548 (23.7%) visits. Both a parenteral and an oral antimicrobial formulation were administered in 3,393 (61.2%) of those visits.
Frequency of antimicrobial use
The most frequently used antimicrobial classes and combinations of classes within the 3 evaluated hospital services by species were summarized (Figure 4). Within all 3 services, penicillins were the most commonly used antimicrobial class, dispensed in 5,628 of the 10,081 (55.8%) visits in which an antimicrobial was dispensed. In 1,339 (13.3%) of these visits, penicillin was the only antimicrobial dispensed. Penicillins and aminoglycosides were used together most commonly in equids, whereas penicillins and cephalosporins were used most commonly in ruminants (bovids, sheep, and goats) and penicillins and sulfonamides were used most commonly in other species (swine, camelids, and cervids).
Proportion of visits (n = 9,436) in which the 5 most commonly used antimicrobials or combinations of antimicrobials were dispensed for the equids (A), ruminants (ie, bovids, sheep, and goats; B), and other species (ie, camelids, swine, and cervids; C) of Figure 1, by hospital service. Amino = Aminoglycosides. Ceph = Cephalosporins. Macro = Macrolides. Pen = Penicillins. Phen = Phenicols. Quin = Quinolones. Sulfa = Sulfonamides. Tetra = Tetracyclines. See Figure 2 for remainder of key.
Citation: American Journal of Veterinary Research 81, 2; 10.2460/ajvr.81.2.103
Discussion
In the present study, antimicrobial prescribing practices differed significantly by the attending hospital service, and substantial heterogeneity in these practices was identified among attending clinicians, even among clinicians within a service for animals with disease affecting the same body system. Heterogeneity in the proportion of visits in which an antimicrobial was dispensed was slightly lower among clinicians within the Surgery service for patients with gastrointestinal disease, but not for any of the other services, nor for patients with musculoskeletal disease. Across the entire hospital, antimicrobials and CIAHPs were dispensed in 42% and 24% of visits, respectively. A median of 3.6 ADDs and a mean of 2.0 classes or formulations of antimicrobials were dispensed per patient, and across all services, penicillins were the most commonly used antimicrobial class. In general, fewer antimicrobials were dispensed and antimicrobials were dispensed less frequently for equids than for other species, whereas more antimicrobials were dispensed and antimicrobials were dispensed more frequently for animals with integumentary and respiratory signs. Also, in general, fewer antimicrobials were dispensed and antimicrobials were dispensed less frequently for older patients as well, whereas the opposite pattern was observed for increasing duration of hospitalization. Finally, submission (vs no submission) of specimens for bacterial culture was associated with antimicrobials (including CIAHPs) being dispensed more frequently, greater numbers of ADDs per visit, and higher numbers of antimicrobials dispensed per visit.
The differences in prescribing patterns among clinicians across the 3 hospital services with the highest caseloads likely reflected the different patient populations of those services. However, these differences could have also reflected the culture of the service, a phenomenon that has been demonstrated in human medicine.45 The significant heterogeneity in prescribing patterns among clinicians was not surprising; similar heterogeneity in such patterns has been shown among physicians.46–48 Our regression models identified patient- and visit-level factors that were significantly associated with whether an antimicrobial or CIAHP was dispensed as well as the number of antimicrobials and ADDs dispensed per visit. In our hospital, patient populations could have differed by clinician; some clinicians may have attended to only food animals, whereas others may have almost exclusively attended to racehorses; some clinicians may have mostly attended to neonates, whereas others may have attended to patients of all ages. Heterogeneity in the patient population can be associated with heterogeneity in indications and options for treatment (eg, some antimicrobials cannot be used in food animals or animal owners may differ in their ability to afford certain treatments). The decrease in interclinician heterogeneity observed in the proportion of visits in which an antimicrobial was dispensed for patients with gastrointestinal disease within the Surgery service supported the hypothesis that the body system affected (a proxy for indication for treatment) may influence antimicrobial prescribing practices in large animal medicine.
Although some variation existed in patient populations attended to by different clinicians, and whereas our regression modeling results suggested that patient factors were significantly associated with antimicrobial prescribing patterns, not all of the variability in prescribing patterns was explained by the factors included in our models, suggesting that other factors may have influenced prescribing patterns of individual clinicians. The consistent heterogeneity in the proportions of visits in which an antimicrobial was dispensed and in the number of ADDs per visit among clinicians attending to patients with musculoskeletal disease supported this hypothesis. One study46 demonstrated that antimicrobial prescribing patterns of individual physicians differ dramatically, even when accounting for characteristics of the patients seen by the physician, and many studies12,30,32,45,49–52 in both human and veterinary medicine have shown that antimicrobial prescribing is a complex phenomenon governed by sociological determinants such as perceptions of professional responsibilities, perceived patient expectations, peer approval, hierarchy within the hospital, and “prescribing etiquette.” These factors are not easily measurable, particularly in a database of administrative records. More research is therefore needed to understand the drivers of antimicrobial prescribing for individual clinicians, especially for specific disease conditions.
Heterogeneity in antimicrobial prescribing patterns and numbers of ADDs prescribed per patient among clinicians for specific clinical conditions can reflect disagreements in how antimicrobials should be prescribed. These disagreements are understandable given the dearth of antimicrobial use guidelines for specific disease conditions in large animals, particularly regarding duration of use.7 However, this was unlikely to have been the sole cause of the heterogeneity observed in our study. For example, in our hospital, the practice of using antimicrobials perioperatively for certain surgical procedures (eg, uncomplicated elective arthroscopic procedures) differed by the attending clinician. Although a previous study53 revealed no differences in the rate of severe complications for horses that did or did not perioperatively receive antimicrobials for clean arthroscopic procedures, more compelling studies are needed to provide evidence to inform hospital policy or change individual prescribing practices, particularly if these prescribing practices are entrenched by years in clinical practice.45 Once evidence-based use guidelines for specific disease conditions are developed, hospital policy and compliance requirements might need to be developed to standardize antimicrobial use. For example, a recent assessment of prescribing practices related to the perioperative use of antimicrobials at our hospital led to the implementation of a formal protocol to standardize the dosing and timing of perioperative antimicrobial treatment.8 Nevertheless, more studies are needed to provide evidence on appropriate antimicrobial treatment for common large animal diseases.
Although antimicrobial use guidelines for specific disease conditions in large animals are lacking, broad guidelines applicable to veterinary patients in general are available.7 For example, performance of bacterial culture followed by antimicrobial susceptibility testing of isolates is recommended to guide prescribing. In the study reported here, submission of specimens for bacterial culture was significantly associated with an increase in the number of antimicrobials dispensed per patient (ie, a greater number of different types of antimicrobials prescribed per patient), suggesting that at least in some situations, antimicrobial susceptibility testing might have been used appropriately to guide a switch in antimicrobial prescribing. However, there were also instances in which antimicrobial treatment was changed without performance of bacterial culture or antimicrobial susceptibility testing. Switching antimicrobials without a compelling reason to do so (ie, identification of the responsible microorganism, lack of clinical response, antimicrobial susceptibility testing results, or allergic reaction) could represent inappropriate antimicrobial use. More information would be needed to determine which factors governed the switch from one antimicrobial to another.
Another broad recommendation for the judicious use of antimicrobials in veterinary medicine is to transition patients as early as possible from parenteral to oral administration.7 In our study, in 61.2% of visits in which 2 or more classes of antimicrobials were dispensed, both a parenteral and an oral antimicrobial formulation were administered, suggesting the transition from parenteral to oral antimicrobial administration occured as recommended. However, because we only had information on total amounts dispensed and no information regarding the temporality of prescriptions, this presumption remains to be verified.
The preferred use of antimicrobials with a narrow spectrum of activity is a tenet of judicious antimicrobial use.7,54 In the study reported here, penicillins, which have a narrow spectrum,55 were the most commonly used antimicrobial class across all services and species. However, this finding did not necessarily represent a stewardship accomplishment, given that penicillins were used as a monotherapy in only 13.3% of the visits in which an antimicrobial was dispensed; much more frequently, they were dispensed in combination with another antimicrobial (eg, an aminoglycoside in horses or third generation cephalosporin in food animals), presumably to achieve a broader spectrum of antimicrobial activity. The combination of penicillin and a third generation cephalosporin, which was the second most commonly used combination, may represent an example of unnecessary antimicrobial use because penicillin does not usually provide any additional coverage over that achieved with ceftiofur.55 Education of veterinary clinicians on the types of bacteria for which the various antimicrobials have efficacy and institution of hospital protocols to guide antimicrobial administration for given pathogens and disease conditions would be fairly simple interventions that could improve antimicrobial prescribing practices.
The use of CIAHPs in veterinary medicine is controversial, although it has been recommended that these antimicrobials be used only when no alternative antimicrobials have been authorized for the respective target bacterial species and disease indication56 and when diagnostic test results support their use.57 One study58 of antimicrobial prescription data from equine practices showed that 3 CIAHPs (enrofloxacin, clarithromycin, and ceftiofur) were included in 7.5% of all prescriptions. In our hospital, the proportion of prescriptions involving a CIAHP was generally low (24.0%). However, some clinicians used CIAHPs frequently (up to 100% of the time), suggesting that interventions aimed at educating clinicians about limiting the use of these drugs may be useful. Under certain circumstances, formulary restrictions could even be considered, although restriction-based stewardship initiatives have been associated with opposition among prescribing physicians and attempts to avoid triggering restrictions (eg, waiting until after hours to prescribe a CIAHP).59,60
As has been shown by other investigators,61 the use of different metrics of antimicrobial use can result in different rankings of antimicrobial use at the hospital service, species, or clinician level. For example, we found that the Emergency and Critical Care service generally used antimicrobials the most frequently of all evaluated services, in addition to using the largest number and quantity of antimicrobials, but rankings for the other 2 services differed with the metric: the percentage of visits in which an antimicrobial was dispensed was higher for the Medicine and Ophthalmology service than for the Surgery service (45.7% vs 43.6%, respectively), but the opposite was observed regarding the median number of ADDs per patient (2.3 vs 3.3). Similarly, there was no consistency in the ranking of clinicians in their prescribing practices across metrics. With the understanding that prescribing fewer antimicrobials does not necessarily mean prescribing more judiciously, ranking of clinicians relative to other clinicians who attend to similar patient populations has been a successful component of antimicrobial stewardship initiatives in human medicine62,63 and is an important first step in understanding why prescribing patterns differ among clinicians. Additionally, although correlating antimicrobial prescribing practices with clinical outcomes was beyond the scope of the present study (and would generally be better suited for randomized clinical trials than retrospective observational studies), our study provided an important foundation for future studies to perform this type of analysis.
The present study had several limitations. Perhaps the most important was that we could not properly account for the indications for treatment or the severity of disease, although an attempt was made to control for duration of hospitalization. Some of the coded diagnoses enabled us to categorize the primary diagnosis by affected body system; however, these data were missing for 12.5% of visits, and patients often had multiple diagnoses, which meant that we did not definitively know for which condition the antimicrobial was prescribed. Our findings did confirm the persistence of heterogeneity in clinician prescribing patterns even for animals with signs affecting the same body system.
Another limitation was that the assumptions underlying the calculations of the number of ADDs associated with a visit could have yielded imprecise results. Data on body weight were missing from the administrative record for one-third of visits, and an estimated weight was assigned instead, which could have resulted in over- or underestimation of the number of ADDs. Moreover, in some instances, the amount of antimicrobial dispensed was greater than the amount administered, particularly for smaller species. For example, in our hospital at the time of the study, ceftiofur was dispensed in a 1-g vial, and although a small ruminant or pig would receive far less than the contents of the entire vial, the amount dispensed (and used in the ADD calculation) would have been 1 g, resulting in overestimation. In addition, because the New Bolton Center hospital is a tertiary care center, the study findings may not be generalizable to primary care situations, in which patients with less severe illness and less need for broader spectrum antimicrobials might be expected.
In conclusion, antimicrobial prescribing patterns in the present study were highly heterogeneous across the 3 hospital services with the highest caseloads and among veterinary clinicians within a service in terms of frequency, quantity, and distribution of use. Submission of specimens for bacterial culture and patient species, age, affected body system, and duration of hospitalization were all significantly associated with the likelihood of having an antimicrobial dispensed and with the amount of antimicrobial dispensed. However, even when these factors were controlled for, a large amount of heterogeneity in prescribing practices remained, even among clinicians within a service who attended to similar patient populations or animals with clinical signs related to the same body system. This suggested that clinician habits and preferences are strong drivers of prescribing practices, which can have important implications for antimicrobial stewardship, treatment outcomes, and the rate of development of antimicrobial resistance. More research is needed to examine heterogeneity in prescribing practices for specific disease conditions—particularly conditions for which antimicrobial use guidelines are lacking—to establish and validate methodologies to assess appropriateness of use and to establish a framework of data collection to provide evidence through which formal antimicrobial stewardship programs could be established and assessed.
Acknowledgments
This study was performed at the University of Pennsylvania School of Veterinary Medicine.
Funded by the University of Pennsylvania.
The authors declare that there were no conflicts of interest.
ABBREVIATIONS
| ADD | Animal-defined daily dose |
| CIAHP | Critically important antimicrobial of the highest priority |
| DDDvet | Veterinary-defined daily dose |
| IQR | Interquartile (25th to 75th percentile) range |
| IRR | Incidence rate ratio |
Footnotes
Stata, version 15, StataCorp, College Station, Tex.
References
1. Singleton DA, Sanchez-Vizcaino F, Arsevska E, et al. New approaches to pharmacosurveillance for monitoring prescription frequency, diversity, and co-prescription in a large sentinel network of companion animal veterinary practices in the United Kingdom, 2014–2016. Prev Vet Med 2018;159:153–161.
2. Medina E, Pieper DH. Tackling threats and future problems of multidrug-resistant bacteria. Curr Top Microbiol Immunol 2016;398:3–33.
3. CDC. Antibiotic resistance threats in the United States, 2013. Available at: www.cdc.gov/drugresistance/pdf/ar-threats-2013-508.pdf. Accessed Aug 14, 2019.
4. Song SJ, Lauber C, Costello EK, et al. Cohabiting family members share microbiota with one another and with their dogs. eLife 2013;2:e00458.
5. van Duijkeren E, Wolfhagen MJ, Box AT, et al. Human-to-dog transmission of methicillin-resistant Staphylococcus aureus. Emerg Infect Dis 2004;10:2235–2237.
6. Cefai C, Ashurst S, Owens C. Human carriage of methicillin-resistant Staphylococcus aureus linked with pet dog. Lancet 1994;344:539–540.
7. Weese JS, Giguère S, Guardabassi L, et al. ACVIM consensus statement on therapeutic antimicrobial use in animals and antimicrobial resistance. J Vet Intern Med 2015;29:487–498.
8. Dallap-Schaer BL, Linton JK, Aceto H. Perioperative antimicrobial drug use and post-operative complications in 762 surgical equine colic patients. J Vet Intern Med 2012;26:1449–1456.
9. Black DM, Rankin SC, King LG. Antimicrobial therapy and aerobic bacteriologic culture patterns in canine intensive care unit patients: 74 dogs (January-June 2006). J Vet Emerg Crit Care (San Antonio) 2009;19:489–495.
10. Murphy CP, Reid-Smith RJ, Boerlin P, et al. Out-patient antimicrobial drug use in dogs and cats for new disease events from community companion animal practices in Ontario. Can Vet J 2012;53:291–298.
11. Summers JF, Hendricks A, Brodbelt DC. Prescribing practices of primary-care veterinary practitioners in dogs diagnosed with bacterial pyoderma. BMC Vet Res 2014;10:240.
12. Speksnijder DC, Jaarsma AD, van der Gugten AC, et al. Determinants associated with veterinary antimicrobial prescribing in farm animals in the Netherlands: a qualitative study. Zoonoses Public Health 2015;62(suppl 1):39–51.
13. Banfield Pet Hospital. Are we doing our part to prevent superbugs? Antimicrobial usage patterns among companion animal veterinarians. Vancouver, Wash: Banfield Pet Hospital, 2017.
14. Pleydell EJ, Souphavanh K, Hill KE, et al. Descriptive epidemiological study of the use of antimicrobial drugs by companion animal veterinarians in New Zealand. N Z Vet J 2012;60:115–122.
15. Wayne A, McCarthy R, Lindenmayer J. Therapeutic antibiotic use patterns in dogs: observations from a veterinary teaching hospital. J Small Anim Pract 2011;52:310–318.
16. Visschers VH, Backhans A, Collineau L, et al. A comparison of pig farmers' and veterinarians' perceptions and intentions to reduce antimicrobial usage in six European countries. Zoonoses Public Health 2016;63:534–544.
17. Jacob ME, Hoppin JA, Steers N, et al. Opinions of clinical veterinarians at a US veterinary teaching hospital regarding antimicrobial use and antimicrobial-resistant infections. J Am Vet Med Assoc 2015;247:938–944.
18. AVMA Task Force for Antimicrobial Stewardship in Companion Animal Practice. Understanding companion animal practitioners' attitudes toward antimicrobial stewardship. J Am Vet Med Assoc 2015;247:883–884.
19. Banfield Pet Hospital. A feline focus on antimicrobial usage. Vancouver, Wash: Banfield Pet Hospital, 2018.
20. Weese JS, Blondeau JM, Boothe D, et al. Antimicrobial use guidelines for treatment of urinary tract disease in dogs and cats: antimicrobial guidelines working group of the International Society for Companion Animal Infectious Diseases. Vet Med Int 2011;2011:263768.
21. Lappin MR, Blondeau J, Boothe D, et al. Antimicrobial use guidelines for treatment of respiratory tract disease in dogs and cats: Antimicrobial Guidelines Working Group of the International Society for Companion Animal Infectious Diseases. J Vet Intern Med 2017;31:279–294.
22. Fowler H, Davis MA, Perkins A, et al. A survey of veterinary antimicrobial prescribing practices, Washington State 2015. Vet Rec 2016;179:651.
23. Weese JS. Investigation of antimicrobial use and the impact of antimicrobial use guidelines in a small animal veterinary teaching hospital: 1995–2004. J Am Vet Med Assoc 2006;228:553–558.
24. Hughes LA, Pinchbeck G, Callaby R, et al. Antimicrobial prescribing practice in UK equine veterinary practice. Equine Vet J 2013;45:141–147.
25. Hardefeldt LY, Browning GF, Thursky K, et al. Antimicrobials used for surgical prophylaxis by companion animal veterinarians in Australia. Vet Microbiol 2017;203:301–307.
26. Gibbons JF, Boland F, Buckley JF, et al. Influences on antimicrobial prescribing behaviour of veterinary practitioners in cattle practice in Ireland. Vet Rec 2013;172:14.
27. AVMA. Antimicrobials: guidelines for judicious therapeutic use. Available at: www.avma.org/KB/Resources/Reference/Pages/Antimicrobial-use.aspx. Accessed Aug 14, 2019.
28. Solomon DH, Van Houten L, Glynn RJ, et al. Academic detailing to improve use of broad-spectrum antibiotics at an academic medical center. Arch Intern Med 2001;161:1897–1902.
29. Lesprit P, Landelle C, Brun-Buisson C. Clinical impact of unsolicited post-prescription antibiotic review in surgical and medical wards: a randomized controlled trial. Clin Microbiol Infect 2013;19:E91–E97.
30. Mateus AL, Brodbelt DC, Barber N, et al. Qualitative study of factors associated with antimicrobial usage in seven small animal veterinary practices in the UK. Prev Vet Med 2014;117:68–78.
31. Coyne LA, Latham SM, Dawson S, et al. Antimicrobial use practices, attitudes and responsibilities in UK farm animal veterinary surgeons. Prev Vet Med 2018;161:115–126.
32. Coyne LA, Latham SM, Williams NJ, et al. Understanding the culture of antimicrobial prescribing in agriculture: a qualitative study of UK pig veterinary surgeons. J Antimicrob Chemother 2016;71:3300–3312.
33. De Briyne N, Atkinson J, Pokludova L, et al. Factors influencing antibiotic prescribing habits and use of sensitivity testing amongst veterinarians in Europe. Vet Rec 2013;173:475.
34. Schwechler J, van den Hoven R, Schoster A. Antimicrobial prescribing practices by Swiss, German and Austrian equine practitioners. Vet Rec 2016;178:216.
35. Hopman NEM, Hulscher M, Graveland H, et al. Factors influencing antimicrobial prescribing by Dutch companion animal veterinarians: a qualitative study. Prev Vet Med 2018;158:106–113.
36. Hughes LA, Williams N, Clegg P, et al. Cross-sectional survey of antimicrobial prescribing patterns in UK small animal veterinary practice. Prev Vet Med 2012;104:309–316.
37. Van Cleven A, Sarrazin S, de Rooster H, et al. Antimicrobial prescribing behaviour in dogs and cats by Belgian veterinarians. Vet Rec 2018;182:324.
38. World Health Organization. Critically important antimicrobials for human medicine. Geneva: World Health Organization, 2013;1–41.
39. McKiernan B. Estimating a horse's weight. Primefact report 494. Orange, Australia: NSW Department of Primary Industries, 2007.
40. Hendricks BL. International encyclopedia of horse breeds. Norman, Okla: University of Oklahoma Press, 2007.
41. Jensen VF, Jacobsen E, Bager F. Veterinary antimicrobial-usage statistics based on standardized measures of dosage. Prev Vet Med 2004;64:201–215.
42. Heinrichs AJ, Wells SJ, Hurd HS, et al. The National Dairy Heifer Evaluation Project: a profile of heifer management practices in the United States. J Dairy Sci 1994;77:1548–1555.
43. Gaddour A, Najari S. Adjustment of the kid's gowth curve in pure goat breeds and crosses under southern Tunisian conditions. J Appl Anim Res 2008;34:117–120.
44. Brown-Douglas CG, Pagan JD. Body weight, wither height and growth rates in Thoroughbreds raised in America, England, Australia, New Zealand and India. In: Pagan JD, ed. Advances in equine nutrition IV. Nottingham, England: Nottingham University Press, 2009;213–220.
45. Charani E, Castro-Sanchez E, Sevdalis N, et al. Understanding the determinants of antimicrobial prescribing within hospitals: the role of “prescribing etiquette”. Clin Infect Dis 2013;57:188–196.
46. Jung S, Sexton ME, Owens S, et al. Variability of antibiotic prescribing in a large healthcare network despite adjusting for patient-mix: reconsidering targets for improved prescribing. Open Forum Infect Dis 2019;6:ofz018.
47. Jones BE, Sauer B, Jones MM, et al. Variation in outpatient antibiotic prescribing for acute respiratory infections in the veteran population: a cross-sectional study. Ann Intern Med 2015;163:73–80.
48. Gerber JS, Prasad PA, Russell Localio A, et al. Variation in antibiotic prescribing across a pediatric primary care network. J Pediatric Infect Dis Soc 2015;4:297–304.
49. Papoutsi C, Mattick K, Pearson M, et al. Social and professional influences on antimicrobial prescribing for doctors-in-training: a realist review. J Antimicrob Chemother 2017;72:2418–2430.
50. Mattick K, Kelly N, Rees C. A window into the lives of junior doctors: narrative interviews exploring antimicrobial prescribing experiences. J Antimicrob Chemother 2014;69:2274–2283.
51. Szymczak JE, Newland JG. The social determinants of antibiotic prescribing: implications for the development and implementation of stewardship interventions. In: Barlam TF, Neuhauser MM, Tamma PD, et al, eds. Practical implementation of an antibiotic stewardship program. Cambridge, England: Cambridge University Press, 2018, 45–62.
52. Mangione-Smith R, McGlynn EA, Elliott MN, et al. The relationship between perceived parental expectations and pediatrician antimicrobial prescribing behavior. Pediatrics 1999;103:711–718.
53. Stöckle SD, Failing K, Koene M, et al. Postoperative complications in equine elective, clean orthopaedic surgery with/without antibiotic prophylaxis. Tierarztl Prax Ausg G Grosstiere Nutztiere 2018;46:81–86.
54. Barlam TF, Cosgrove SE, Abbo LM, et al. Implementing an antibiotic stewardship program: guidelines by the Infectious Diseases Society of America and the Society for Healthcare Epidemiology of America. Clin Infect Dis 2016;62:e51–e77.
55. Bush K. Beta-lactam antibiotics: penicillins. In: Finch RG, Greenwood D, Whitley RJ, et al, eds. Antibiotic and chemotherapy. 9th ed. Philadelphia: WB Saunders Co, 2010;200–225.
56. European Medicines Agency. Answers to the requests for scientific advice on the impact on public health and animal health of the use of antibiotics in animals. EMA/381884/2014. London: European Medicines Agency, 2014.
57. Limb M. Protecting critically important antimicrobials. Vet Rec 2018;182:651.
58. Welsh CE, Parkin TDH, Marshall JF. Use of large-scale veterinary data for the investigation of antimicrobial prescribing practices in equine medicine. Equine Vet J 2017;49:425–432.
59. Drew RH, White R, MacDougall C, et al. Insights from the Society of Infectious Diseases Pharmacists on antimicrobial stewardship guidelines from the Infectious Diseases Society of America and the Society for Healthcare Epidemiology of America. Pharmacotherapy 2009;29:593–607.
60. LaRosa LA, Fishman NO, Lautenbach E, et al. Evaluation of antimicrobial therapy orders circumventing an antimicrobial stewardship program: investigating the strategy of “stealth dosing.” Infect Control Hosp Epidemiol 2007;28:551–556.
61. Mills HL, Turner A, Morgans L, et al. Evaluation of metrics for benchmarking antimicrobial use in the UK dairy industry. Vet Rec 2018;182:379.
62. Hamilton KW, Fishman NO. Antimicrobial stewardship interventions: thinking inside and outside the box. Infect Dis Clin North Am 2014;28:301–313.
63. Hamilton KW, Gerber JS, Moehring R, et al. Point-of-prescription interventions to improve antimicrobial stewardship. Clin Infect Dis 2015;60:1252–1258.
Appendix
Classification of antimicrobials used at the New Bolton Center hospital of the University of Pennsylvania from 2013 through 2018.
| Drug class | Active ingredients | Formulation |
|---|---|---|
| Cephalosporins | Cefazolin, cefotaxime,* ceftiofur,* ceftriaxone,* and cefuroxime | I, OR, OP, and IMAM |
| Penicillins | Penicillin, amoxicillin, ampicillin, and penicillamine | I, OR, and IMAM |
| Tetracyclines | Oxytetracycline, doxycycline, and minocycline | I, OR, and OP |
| Aminoglycosides | Amikacin, gentamicin, tobramycin, and neomycin | I, OR, OP, and T |
| Quinolones | Enrofloxacin,* ciprofloxacin,* and ofloxacin* | I, OR, and OP |
| Macrolides | Azithromycin,* clarithromycin,* erythromycin,* gamithromycin,* and tulathromycin* | I, OR, and OP |
| Phenicols | Chloramphenicol and florfenicol | I, OR, and OP |
| Sulfonamides | Sulfadimethoxine, sulfadiazine, sulfadoxine-pyrimethamine, and trimethoprim-sulfamethoxazole | I, OR, and T |
| Polypeptides | Bacitracin and polymyxin* | I and OP |
| Other | Rifampin, pirlimycin, metronidazole, nitrofurazone, and meropenem | I, IMAM, OP, OR, and T |
A CIAHP according to the World Health Organization.
I = Injectable. IMAM = Intramammary. OP = Ophthalmic. OR = Oral. T = Topical.
