A double-blinded randomised clinical trial was conducted between the months of July and October of 2014. The experimental protocol was approved by the Comité d’éthique et d’utilisation des animaux de l’Université de Montréal (protocol: 14-Rech_1727).
Population and study design
The study was performed in one commercial pre-weaning veal calf fattening unit in St-Hélène de Bagot, Québec (Délimax, Veaux Lourds Ltée, Saint-Hyacinthe, Québec, Canada). Calves coming from local auction markets were enrolled at the approximate age of seven days of life (exact age was not reported for each calf) on arrival at the farm. A sample size estimation of 200 calves (100 calves per treatment group) was based on detecting a difference in treated versus control calves of 4 kg of weight at the end of the pre-weaning period (105 kg in control vs 109 kg in treated calves; assuming a population variance (s2) of 100) as well as a difference of 15% in the proportion of calves treated for BRD (25% in control vs 10% or less in treated calves) with a statistical power of 80% and an alpha error of 5%. The difference of 4 kg was defined as the minimal difference financially beneficial for the producer considering the cost of tildipirosin. The proportions used in this study were estimated based on previous data of BRD reported in veal calves by the owner and the literature [1], and the differences were based on those observed in feedlots [11].
On arrival at the farm (day 0, D0), calves were immediately intra-nasally vaccinated against infectious bovine rhinotracheitis and para-influenza 3 virus (Nasalgen® IP, Merck Animal Health, Kirkland, Québec, Canada) and were housed in duckboard individual pens and fed with milk replacer (21% crude protein, 18.5% fat) until a quantity of 25 to 28 kg at the time of weaning period 70 days after arrival. Each calf was identified using an ear tag, and calf pen allocation was also recorded. Calves were randomly distributed between two groups using the RAND function in Excel (Microsoft, Redmond, WA, USA) software. One group of calves received tildipirosin (TILD) at the labelled dose subcutaneously (SQ) (4 mg/kg). The second group received an equivalent volume of 0.9% saline SQ (PLAC). The metaphylactic antimicrobial and placebo treatments were given at day 12 (D12), which was one week before the main peak of clinical incidence according to historical health observations in this fattening unit (Annie Dubuc, Dr. Frédéric Beaulac, Délimax veaux lourds Ltée, personal communication) that is similar of the peak reported in veal calves in the litterature [1].
The commercial solution of tildipirosin (18%) and sterile saline (0.9%) were specifically prepared for the trial, using opaque vials, and they were kept locked away by a pharmacist. The bottles were labelled with either an A or a B, and the corresponding keys were kept by the pharmacist during the entire study period. The bottle identifications were only revealed to the researcher at the end of the data collection period.
Data collection
The different data collected and their times of collection are presented in Fig. 1.
The first visit was performed by a veterinarian and an animal health technician one day after arrival (D1). Calves were bled from the jugular vein for assessment of serum total solids. Blood samples were centrifuged at 3500 rpm for 10 min at room temperature. The serum total solids were then determined using a manual handheld refractometer (Atago SUR-NE, Tokyo, Japan). Failure of passive transfer (FPT) was defined when total solids were less than or equal to 52 g/l [13].
Throughout the study period, the producer was allowed to treat the calves according to on-farm protocols but did not participate in the research project visits. Individual medical treatments or events, with a focus on BRD treatments, and mortalities were recorded during the entire pre-weaning period but no necropsies were performed. The number of BRD treatments was used to estimate the apparent prevalence by the producer.
On D1, D12 (the day of metaphylactic treatment) and day 30 (D30), a respiratory score (RS) was also performed on each calf using the Calf Respiratory Scoring Chart [14] as previously described [15]. An RS of five or more has been suggested as a BRD clinical case [14]. All clinical examinations were performed by two different operators blinded to TUS results and treatment allocation.
On D1, D12 and D30, TUS was performed with a 7.5 MHz linear probe (Imago, ECM, Angoulème, France) on the right and left thorax by two different operators blinded to respiratory score results and treatment allocation. The area screened was the mid to ventral portion of the right and left lung as well as the parenchyma cranial to the heart [16]. Thoracic ultrasonography was used to detect calves with lung consolidation. The maximal depth of consolidation (DEPTH) was recorded. The DEPTH (cm) was calculated by a manual count of the 1 cm squares using the 1 cm2 grid of the ultrasound unit. As previously described, lung consolidation was considered significant when at least one point of DEPTH greater than or equal to 3 cm was observed [16].
During the study period, the calves were weighed by a blind research assistant three times: at arrival (D0, W0), at D30 (W2) and just before leaving the fattening unit at day 70 (W3). Individual calf weights were determined by a digital scale. Average daily gain was defined for the first month of feeding (ADG1; difference between W2 and W0 divided by 30 days), for the second month of feeding (ADG2; difference between W3 and W2 divided by 40 days) and for the total of the pre-weaning period (ADG tot; difference between W3 and W0 divided by 70 days).
Statistical analyses
The calf was the unit of interest in this study. As recommended by Sargeant [17], intention-to-treat (ITT) analyses were used. This analysis compares the group exactly as randomised. This approach is considered the gold standard if there are losses to follow-up in a randomised controlled trial because randomisation is maintained and adverse effects, as well as lack of compliance, is likely to occur in “real world” use of the intervention [17]. For the only calf that died before the third visit at D30, the medians of ultrasonographic lung consolidation, of clinical score and of ADG, were assigned [18]. For the calves that died after D30, a last observation carried forward approach [18] was performed for ADG, i.e. the ADG during the second period of the dead calves (ADG2) was considered similar to the one observed during the first period (ADG1; ADG1 = ADG2). Two calves had no weight at D30 to calculate ADG1 and ADG2, and ADG tot was attributed to calves.
Statistical analyses were performed using SAS (version 9.3, SAS Institute Inc. Cary, NC, USA). Descriptive statistics were first calculated for the two treatment groups using the MEANS and FREQ procedures in SAS. A T-test was used to compare both treatment groups concerning the continuous variables (total solids at arrival, weight, ADG), with statistical significance set at P < 0.05. Fisher’s exact test was used to compare both treatment groups concerning the dichotomized variables (breed, sex, FPT, TUS, RS, mortality, BRD treatments by the producer after administration of TILD), with statistical significance set at P < 0.05.
Secondly, logistic regression analyses were performed to assess the impact of treatment group on the presence of significant lung consolidation on D30 (GLIMMIX procedure with LOGIT link in SAS). Univariable model building was performed using Chi-squared or Fisher’s exact tests (FREQ procedure). The covariates with P < 0.25 were retained for building a final multivariable linear regression model using a backward elimination strategy. The final model was obtained when all covariates had P < 0.05.
Finally, ADG1, ADG2 and ADG tot were considered as dependent variables in univariable linear regression models (MIXED procedure in SAS). All the potential covariates were first assessed using univariable analyses with ADG as the dependent variable. The covariates with P < 0.25 were retained for building a final multivariable linear regression model using a backward elimination strategy. The final model was obtained when all covariates had P < 0.05. The treatment group was forced in every multivariable model since it was the main study interest.