Analyzed data were from 281,543 consultations for sick dogs (155,732 unique dogs) and 111,139 sick cats (69,236 unique cats) from 173 veterinary practices (379 sites). A descriptive population summary is shown in Table 1, and a summary of genetic breed groups included in this study is shown in Appendix Table 2.

Dogs

Prescription of Antimicrobial Drugs

Systemic antimicrobial drugs were prescribed during 25.7% (95% CI 24.9%–26.6%) of consultations, topical antimicrobials during 14.2% (95% CI 13.9%–14.6%), and systemic HPCIAs during 1.4% (95% CI 1.2%–1.6%). The most commonly prescribed class of systemic HPCIAs was fluoroquinolones (0.9% of sick animal consultations, 95% CI 0.7%–1.0%), followed by third-generation cephalosporins (0.5%, 95% CI 0.4%–0.6%) and macrolides (0.1%, 95% CI 0.0%–0.2%). Antimicrobial prescription summarized according to common consultation by breed is shown in Appendix Table 3.

Prescription of Systemic Antimicrobial Drugs

Figure 1

Figure 1. Results from 2 multivariable mixed effect logistic regression models predicting probability of systemic antimicrobial prescription in study of factors associated with prescription of antimicrobial drugs for dogs and cats, United Kingdom,...

Descriptive analyses and univariable model results are summarized in Appendix Table 4. Final multivariable model results are available in Table 2. Systemic antimicrobial drugs were less likely to be prescribed for vaccinated or neutered dogs than for nonvaccinated or non-neutered dogs. Systemic antimicrobial drugs were also less likely to be prescribed for insured dogs up to ≈12 years of age (Figure 1, panel A). Odds for prescription of a systemic antimicrobial drug were greater for dogs with an MPC that was respiratory than for those with a gastroenteric MPC. Mixed practices were associated with significantly increased odds of this prescription compared with practices treating companion animals only. RCVS-accredited practices were less likely to prescribe a systemic antimicrobial.

Prescription of Systemic HPCIAs

Figure 2

Figure 2. Results from 2 multivariable mixed effect logistic regression models predicting probability of systemic highest priority critically important antimicrobial (HPCIA) prescription in study of factors associated with prescription of antimicrobial drugs for...

Descriptive analyses and univariable model results are summarized in Appendix Table 5. Final multivariable model results are available in Table 3. Systemic HPCIAs were less likely to be prescribed for vaccinated or insured dogs; highest odds for prescription were for dogs with a respiratory MPC. Odds increased with age (Figure 2, panel A). In terms of genetic breed, the greatest odds of systemic HPCIA prescription were for the toy breed group, compared with retrievers.

Prescription of Topical Antimicrobial Drugs

Figure 3

Figure 3. Results from 2 multivariable mixed effect logistic regression models predicting probability of topical antimicrobial prescription in study of factors associated with prescription of antimicrobial drugs for dogs and cats, United Kingdom,...

Descriptive analyses and univariable model results are summarized in Appendix Table 6. Final multivariable model results are available in Table 4. Topical antimicrobial drugs were less likely to be prescribed for insured dogs, although odds increased significantly for male, microchipped, or vaccinated dogs. The effect of age varied according to MPC; an MPC of pruritus was generally associated with greatest odds of topical antimicrobial drug prescription throughout life, broadly decreasing with increased age (Figure 3, panel A). Odds of topical antimicrobial drug prescription were lowest for sight hounds compared with retrievers. Practices employing RCVS specialists were less likely to prescribe a topical antimicrobial.

We demonstrated frequent prescription of antimicrobial drugs, including systemic HPCIAs (particularly in cats), in veterinary practices in Great Britain. Considering the importance of HPCIAs in the context of antimicrobial resistance (32), we identified a vital need to learn more about factors potentially driving such prescribing behaviors. We further augmented EHR data with a range of external data sources to identify key client, animal, and practice-related risk factors associated with prescription of systemic and topical antimicrobial and systemic HPCIA drugs; such factors potentially inform key antimicrobial stewardship targets of importance to companion animal practice.

Regarding client care decision-related factors, odds of prescription for systemic antimicrobial and HPCIA drugs were significantly lower for vaccinated dogs and cats, possibly reflecting perceived or actual reduced risk for antimicrobial drug–responsive disease in vaccinated animals. Although most companion animal vaccines target viruses, bacterial infection secondary to vaccine-preventable viral disease has been documented (33). Risk avoidance plays a major role in antimicrobial drug–prescribing practices (12), potentially prompting more frequent prescription for unwell, nonvaccinated animals. We speculate that previous engagement with preventive healthcare services might select for clients more likely to seek veterinary attention earlier or to pursue diagnostic options rather than empirical prescription. Regardless of what might be driving these trends, the O’Neill report recommends that promoting development and use of vaccines and alternatives to antibiotics should form a key component of efforts to curtail human dissemination of antimicrobial resistance (34); our findings suggest that such recommendations should also be considered for companion animals.

Insurance coverage was associated with decreased odds of prescription of systemic and topical antimicrobial drugs, potentially highlighting veterinary practitioners being more likely to seek a wider range of diagnostic options in preference to empirical antibiosis for insured animals. However, insured dogs were also associated with increased odds of prescription of systemic HPCIA drugs. Cost of therapy has been shown to influence choice of antimicrobial agent for companion animals (17), and HPCIA drugs are anecdotally considered a more expensive option than other antimicrobial drugs. Hence, our findings might reflect increased willingness to prescribe relatively expensive antimicrobial drugs for insured dogs.

Although HPCIA drug classification remains under debate, use of HPCIA drugs has formed a focus for antimicrobial resistance–related policy (34). Although several classes of HPCIA drugs (e.g., glycopeptides, which are not authorized for use in animals) are very rarely prescribed for companion animals in the United Kingdom (22), prescription of fluoroquinolones and third-generation cephalosporins (particularly for cats) is relatively common, although current antimicrobial drug prescribing guidance strongly discourages such practices (35).

With regard to animal-intrinsic factors, odds for prescription of systemic antimicrobial drugs were increased for younger male cats, although the opposite was found for dogs. Sex-based variation in risk for bacterial infection has been identified (36–38), and cat fight–related injuries are a frequently recorded clinical complaint (39), more commonly associated with young outdoor-ranging male cats (40). Indeed, we found that systemic antimicrobial drugs were more commonly prescribed for male cats presented for trauma. Furthermore, time of injury is less likely to be known for outdoor-ranging cats than for dogs; such uncertainty might prompt a more cautious approach to prescription of antimicrobial drugs (41).

Other studies have also identified age- or sex-related variation in risk for antimicrobial resistance (36–38). For instance, Radford et al. demonstrated decreased probability of systemic antimicrobial prescription with increased patient age (20), potentially reflecting increased actual or perceived incidence of noncommunicable disease as animals age. This interpretation might partly explain our findings, although we noted an exception with prescription of systemic HPCIA drugs. For cats, an easy-to-administer (injectable) long-acting third-generation cephalosporin formulation is widely used (21–23). Although not completely explanatory, our findings may suggest that as an animal ages, the client or veterinarian perceives an increased probability of an animal being refractory to an intervention (e.g., oral administration of tablets), increasing the likelihood of a prescriber choosing easy-to-administer formulations. A previously identified key influencer of antimicrobial agent choice was administration of inappropriate dosages as a result of noncompliance (17). Whether the risk for antimicrobial resistance posed by a possible underdose of a first-line antimicrobial outweighs the risk posed by the labeled dose of a third-line HPCIA drug remains unanswered.

As with humans (10,11,13), prescription of systemic antimicrobial drugs for dogs and cats was most commonly associated with respiratory clinical signs. Humans having respiratory conditions are often inappropriately prescribed antimicrobial drugs; most respiratory conditions are viral or noninfectious in origin (10). The same has also been shown for companion animals, although bacterial infection secondary to primary viral disease has been documented (42). Considering these shared patterns, although prescribing guidance is available (43), we suggest respiratory disease as a pertinent area for further investigation of One Health antimicrobial stewardship intervention methods.

Increased odds of prescription of topical antimicrobial drugs was commonly associated with the retriever group of dogs, which contains several breeds commonly associated with dermatologic disease (44). This finding and interpretation is plausible, suggesting that the breed summarization technique used here to combat the modeling issues posed by the >250 dog and >50 cat breeds recorded in this dataset was useful. However, genetic linkage does not necessarily imply phenotypic similarity. As such, individual breed-level phenotypes might be responsible for conferring variant bacterial infection risk in ways not explored, and indeed potentially masked. In future analyses, we will aim to identify additional means by which breeds can be effectively summarized according to shared genotype and phenotype.

Although the individual animal accounted for most of the random effect variance seen in this study, veterinary-led factors might well yield more readily accessible routes toward stewardship. For site accreditation, the voluntary RCVS Practice Standards Scheme requires antimicrobial drug use policies, infection control plans, and established clinical audits (45), and we observed reduced systemic antimicrobial prescription odds for dogs in accredited practices. Although practices seeking accreditation might already be more engaged with quality improvement, we nevertheless recommend further consideration as to whether the RCVS Practice Standards Scheme could play a more central role for encouraging stewardship in general and referral practices.

Compared with practices that treat companion animals only, mixed species practices were associated with increased odds of prescription of systemic antimicrobial drugs. Veterinary surgeons employed in different sectors expressed varied attitudes toward antimicrobial resistance (16), a finding perhaps demonstrated on a wide scale in this study. Practices employing RCVS specialists were also associated with reduced odds of prescription of topical antimicrobial drugs for dogs, potentially reflecting varied case management approaches (46) or caseloads compared with general practices.

Considering limitations of this study, although we successfully augmented EHRs with a variety of data sources, no dataset is infallible. For instance, the veterinary surgeon employment record of the RCVS Practice Register is updated only on an ad hoc basis. It is thus possible that the surveyed veterinary surgeon population varied over the 2-year study period in ways not captured here. Veterinary practices participating in SAVSNET are recruited by convenience and might not be representative of the wider UK population. Although we found no clear associations between IMD or pet population density and prescription, the complexities of summarizing IMD across the constituent countries of the United Kingdom (47), coupled with the relative infancy of pet population demographic studies (30), lead us to recommend re-evaluation as research methods further mature. The analyzed population was relatively skewed toward less deprived areas; to ascertain whether this finding is reflective of the wider UK pet-owning community, including the charity and low-income veterinary sectors in future analyses would be warranted. We advise caution for inferring causal relationships between factors and outcome variables explored in this cross-sectional study; similarly, group-level observations might have limited relevance to individual animals. More generalized SAVSNET limitations have been previously discussed; in brief, quantification of antimicrobial drug prescription depends on practitioners charging for antimicrobial drugs, and analyzed practices were recruited by convenience (22,30).

In conclusion, we demonstrated the value of using veterinary EHRs collected from a cohort of veterinary practices to identify a range of factors associated with prescription of antimicrobial drugs for dogs and cats. Although factors influencing decision-making remain multifactorial and complex, our findings suggest that gathering clinical evidence surrounding respiratory disease might be of value to stewardship. Preventive healthcare could also play a role in stewardship and should form the basis of client-targeted health messaging, as should the RCVS Practice Standards Scheme for veterinary practitioners.