ReviewA mixed treatment comparison meta-analysis of antibiotic treatments for bovine respiratory disease
Introduction
Producers and veterinarians may legally choose from numerous registered antibiotics for the treatment of bovine respiratory disease (BRD). When deciding which antibiotic to use producers and veterinarians consider many factors including comparative efficacy. Ideally producers and veterinarians would conduct randomized controlled trials at the production site to compare the efficacy of the products, and make the treatment choice based on observed efficacy and cost. For owners of large feedlots and feedlot consultants this is a realistic approach to selection of treatment regimens. For other veterinarians and producers, conducting clinical trials may be difficult. In these situations, veterinarians and producers rely upon scientific literature to assess the efficacy of products. However, if the comparison of interest is not publically available veterinarians or producers must use indirect information to assess comparative efficacy. For example, if one trial compared regimen B to regimen A and reported a relative risk of failure of 0.5; i.e., B had half the failure rate of A, and a second trial compared regimen C to regimen A and reported a relative risk of failure of 0.25, many would conclude that C was twice as effective as B based on the indirect comparison of B vs. C. This approach to comparative efficacy is referred to as the naïve approach and can be misleading because it ignores study level factors and the unit of randomization (Glenny et al., 2005). The naïve approach also fails to empirically incorporate uncertainty about the within-trial direct estimates. For example, if the B vs. A trial used 1000 animals, while the C vs. A trial used 100 animals, the differences in uncertainty about the direct effects of B vs. A and C vs. A should be incorporated into the estimate of comparative efficacy.
A statistical alternative to the naïve approach for comparative efficacy is called mixed treatment comparison (MTC) meta-analysis, also known as network meta-analysis (Higgins and Whitehead, 1996, Jansen et al., 2008, Lu and Ades, 2004, Lumley, 2002, Salanti et al., 2008). The principle behind MTC meta-analysis is to use evidence from the full network of trial results to make inferences about comparative efficacy while addressing many concerns associated with the naïve approach (Higgins and Whitehead, 1996). Mixed treatment comparison meta-analysis combines direct and indirect estimates of efficacy using the network of information from trials, while accounting for lack of randomization at the study level. The use of mixed treatment comparisons meta-analysis is well established in human medicine but the approach remains rare in veterinary science (Jansen et al., 2011, Nixon et al., 2007, Numthavaj et al., 2011, Piccini and Kong, 2011, Steiner et al., 2012, Van den Bruel et al., 2011, Wang et al., 2010).
Our objective was to use MTC meta-analysis to compare the efficacy of BRD antibiotic treatments. The review question was “What is the comparative efficacy of antibiotics treatments registered for use in North America for the treatment of undifferentiated bovine respiratory disease in feedlot calves?” The rationale for the project was that the information from a MTC meta-analysis would help producers and veterinarians use the full network of publically available information for decision making about BRD treatment choices.
Section snippets
The review question
Review questions are often defined in terms of the population (P), intervention (I), comparator (C) and outcome (O), i.e., a PICO question, and we elaborated each aspect of the question using this approach. The population of interest (P) was calves with undifferentiated bovine respiratory disease in North American feedlots. Therefore, studies were considered eligible if they could reasonably be expected to exclude yearling aged animals. Many studies did not provide exact information about the
Results
The network of evidence used in the analysis contained 194 trial arms from 93 trials (Fig. 1). From 93 trials there were 8 with three arms. (Booker et al., 1997, Freedom of Information Summary, 1996, Freedom of Information Summary, 1988, Freedom of Information Summary, 1998a, Freedom of Information Summary, 1998b, Freedom of Information Summary, 1998c, Freedom of Information Summary, 2002, Freedom of Information Summary, 2003, Freedom of Information Summary, 2005, Freedom of Information
Discussion
The results of the MTC meta-analysis enable producers and veterinarians to obtain estimates of comparative efficacy that better describe uncertainty than naïve approaches to indirect comparisons (Fig. 2). For example, the estimated risk ratio comparing incidence of re-treatment in trials that assessed danofloxacin (numerator) to trial arms that assessed tulathromycin (denominator) is 1.32 (95% CI = 0.66, 2.43) suggesting that based on the publically available data there is no evidence that one
Funding
No direct grant support was used for this project.
Conflict of interest statement
Dr. O’Connor received a grant from Pfizer Animal Health to study an unrelated factor in 2010–2012. The other authors have no conflicts to declare.
Acknowledgements
Thank you to Dr. J. Higgins for addition assistance with the MTC analysis. Thank you to Stacie Gould for assistance with data extraction and manuscript preparation.
References (56)
- et al.
Bayesian meta-analysis of multiple treatment comparisons: an introduction to mixed treatment comparisons
Value Health
(2008) - et al.
Interpreting indirect treatment comparisons and network meta-analysis for health-care decision making: report of the ISPOR Task Force on Indirect Treatment Comparisons Good Research Practices. Part 1
Value Health
(2011) - et al.
The REFLECT statement: methods and processes of creating reporting guidelines for randomized controlled trials for livestock and food safety
Prev. Vet. Med.
(2010) - et al.
Effectiveness of different central venous catheters for catheter-related infections: a network meta-analysis
J. Hosp. Infect.
(2010) - et al.
Evaluation of florfenicol for the treatment of undifferentiated fever in feedlot calves in western Canada
Can. Vet. J.
(1997) - et al.
Checking consistency in mixed treatment comparison meta-analysis
Stat. Med.
(2010) - Freedom of Information Summary, 1988. NADA 140-338 Naxcel Sterile Powder – original approval....
- Freedom of Information Summary, 1996. NADA 141-063 Nuflor Injectable Solution – original approval....
- Freedom of Information Summary, 1998a. NADA 140-890 Excenel Sterile Suspension....
- Freedom of Information Summary, 1998b. NADA 141-063 Nuflor Injectable Solution....
NADA 141-334 Zuprevo 18% – Original Approval
Indirect comparisons of competing interventions
Health Technol. Assess.
Clinical efficacy of enrofloxacin against bovine respiratory disease comparing different treatment regimens
Bovine Pract.
Dose determination and confirmation for ceftiofur crystalline-free acid administered in the posterior aspect of the ear for control and treatment of bovine respiratory disease
Vet. Ther.
Borrowing strength from external trials in a meta-analysis
Stat. Med.
Dose determination and clinical evaluation of the efficacy of enrofloxacin injectable solution in the treatment of bovine respiratory disease
A comparison of the clinical field efficacy and safety of florfenicol and tilmicosin for the treatment of undifferentiated bovine respiratory disease of cattle in western Canada
Can. Vet. J.
A comparison of florfenicol and tilmicosin for the treatment of undifferentiated fever in feedlot calves in western Canada
Can. Vet. J.
Therapeutic efficacy of tulathromycin, a novel triamilide antimicrobial, against bovine respiratory disease in feeder calves
Vet. Ther.
Clinical efficacy of florfenicol in the treatment of bovine respiratory disease in Europe and North America
Cited by (41)
Bovine Respiratory Disease Considerations in Young Dairy Calves
2022, Veterinary Clinics of North America - Food Animal PracticeA mixed treatment meta-analysis of antibiotic treatment options for bovine respiratory disease – An update
2016, Preventive Veterinary MedicineCitation Excerpt :The prior distributions were originally based on the approach reported previously (Dias et al., 2011). For the original model we assessed σ ∼ U(0, 2) and σ ∼ U(0, 5), and that original analysis suggested σ ∼ U(0, 5) was preferred, we retained this prior in the updated model (O’Connor et al., 2013). The fit of the model was assessed based on the log odds ratio, by examining the residual deviance between the predicted values from the MTC model and the observed value for each study (Dias et al., 2010b).
Veterinary Medicine, Eleventh Edition
2016, Veterinary Medicine, Eleventh Edition