Bovine respiratory disease is the most common health problem in feedlot cattle in North America.1,2 The economic impact of BRD on the US beef industry has been reported to exceed $4 billion annually, which includes the costs for treatment, disease prevention, and lost productivity.3 Treatment costs are incurred for calves with BRD during the feedlot phase, but BRD may have longer-lasting effects on performance of affected calves that will impact their final value. To assess potential impacts of disease, evaluation of net returns for individual calves can be used with concurrent standardization for known costs and income structures to provide a framework for evaluating the overall effect of BRD.4 Knowledge of the factors that contribute to differences in net returns could improve the management and marketing plans for specific groups of cattle as well as contribute to development of preventative health programs and strategies for disease mitigation.4 It is challenging to estimate potential disease costs for calves entering feedlots, and an understanding of the actual disease costs could alter management decisions.5 Net returns models can be used to generate an estimate of the actual cost of BRD, which thereby can promote efficient allocation of resources to preventative and treatment programs.
In previous studies,6,7 investigators assessed the economic impact of the frequency of treatments for BRD and BRD health-related outcomes between sick and healthy cattle. In 1 study,7 investigators found a decrease in carcass value of $23.23, $30.15, and $54.01 for cattle treated once, twice, or ≥ 3 times, respectively, compared with the carcass value for cattle that were never treated. However, overall net returns were not estimated.7 Net returns decrease for cattle in the backgrounding and finishing phases as the number of BRD treatments increase.6 Compared with results for chronically ill calves during the backgrounding phase, groups of cattle that were not treated or were treated once, twice, or 3 times typically yielded greater net returns of $111.12, $92.51, $59.98, and $20.62, respectively. Although these studies provided useful information on the effects of BRD on production traits, their estimates were based on small sample sizes and were not generated in a multivariable framework. Estimated net returns are impacted by multiple factors that may also influence disease risk (eg, demographic factors); therefore, use of a multivariable approach would allow a more discrete comparison of the sole effect of the number of treatments on net returns.
Differences in performance depended on when during the feeding phase cattle were treated because of respiratory disease.8,9 Factors that influenced differences in estimated net returns in those studies8,9 included the number of treatments because of BRD, HCW, number of days cattle were fed (ie, DOF) from arrival to slaughter, and ADG. Cattle with a greater interval between treatment because of BRD and slaughter yielded greater estimated net returns related to an increase in HCW that appeared to offset increased costs attributable to more treatments, a higher number of DOF, and a decrease in ADG.9 Yearly fluctuations of cattle, therapeutics, feed, and grid prices inherently affect net returns. Moreover, multiple other performance and health factors can have an adverse impact on the market value of cattle with BRD, which makes cattle costs, sale price, and purchase value less predictable.10 Therefore, an approach that accounts for the frequency of treatments for respiratory disease in addition to demographic and other relevant covariates that may impact net returns is warranted to determine specific factors that affect economic differences observed between sick and healthy cattle.
Our hypothesis was that the number of times cattle are treated during the feeding period after an initial diagnosis of BRD could impact performance and economic outcomes. Thus, the objectives of the study reported here were to evaluate the economic impact (expressed in terms of net returns) of the number of treatments after an initial diagnosis of BRD while accounting for cohort- and individual-level demographic factors in cattle at arrival and to determine whether differences in performance variables (ADG) or carcass traits (HCW, CYG, and QG) of economic importance for commercial feedlot cattle could be explained on the basis of the number of times cattle were treated during the feeding period after an initial diagnosis of BRD and demographic characteristics of the cattle on arrival at a feedlot.
Materials and Methods
Animals—Data for cattle that arrived at a Midwestern feedlot from February 2001 through March 2006 were acquired for the study. Approval of the study by an institutional animal care and use committee was not required because the study did not involve the use of animals; instead we used data collected routinely by feedlot personnel.
Data collection—Data on performance, health, and carcass traits of individual cattle were obtained. The dataset was refined to include, for cattle with BRD, only those cattle in which there was an initial diagnosis (and subsequent treatment) ≤ 100 days after arrival at the feedlot. Records for cattle that died were removed from the dataset because performance data for dead cattle were not available. Records that had incomplete data on morbidity, performance, and treatments because of diseases other than BRD, or for cattle in which BRD was initially diagnosed and treated > 100 days after arrival at the feedlot, were also removed from the analysis.
Cattle procured and managed in a similar manner, but not necessarily housed in the same pens, were defined as a cohort. Most cattle within each cohort were marketed at the same time. A case of BRD was defined as an animal with clinical signs of respiratory tract disease that was identified for the first time by feedlot personnel and subsequently treated with an antimicrobial. The specific health management and selected treatment was determined by the feedlot manager and consulting veterinarian.
Data available on individual cattle included information on demographics (individual identification number, cohort number, arrival date, arrival weight, and sex), health (diagnosis of 1 to 3 episodes of disease), number and type of treatments, BRD morbidity (date or dates and number of times BRD was diagnosed in a calf and that calf was subsequently treated because of BRD during the feeding period), fatalities (date of death), and performance (ADG and DOF). Carcass data (HCW, rib eye muscle area, QG, back fat, longissimus muscle area, yield grade, and kidney, pelvic, and heart fat) were also recorded.
A variable was created pertaining to the number of antimicrobial treatments after an initial diagnosis of BRD. For that variable, zero treatments included cattle for which BRD was never diagnosed and that never received antimicrobial treatment because of BRD or other diseases during the first 100 days after arrival at the feedlot. One treatment included cattle for which there was an initial diagnosis of BRD (subsequent antimicrobial treatment because of BRD), but no additional antimicrobial treatments for other diseases, during the first 100 days after arrival at the feedlot. Two treatments included cattle that received antimicrobial treatments twice because of BRD after an initial diagnosis (treatment after the initial diagnosis of BRD and a subsequent second treatment) during the first 100 days after arrival at the feedlot. Three or more treatments included cattle that received ≥ 3 antimicrobial treatments because of BRD after an initial diagnosis (treatment after the initial diagnosis of BRD and 2 subsequent treatments) during the first 100 days after arrival at the feedlot. Some cattle (n = 1,954) were treated > 3 times. However, there were no data on a specific diagnosis; thus, subsequent treatments could have been administered because of diseases other than BRD. We assumed that cattle receiving ≥ 3 treatments because of BRD had chronic respiratory disease.
Net returns model—The net return for each animal was calculated on the basis of each animal's body weight at the time of arrival, number of DOF, number of treatments, ADG, and carcass characteristics. Net return was calculated by use of the following equation9: net return = base price ± grid price − total expenditures − feeder price. Base price (mean USDA cutout value for boxed beef), grid price (mean USDA values for monthly cattle premiums and discounts for slaughter steers and heifers), and feeder price (mean USDA value determined at Oklahoma City for the monthly feeder calf price aggregated by weight class) were obtained for the period November 1996 through April 2005 to reflect a broad range of market conditions.11 For the time frame of the study, records analysis revealed that total expenditures represented treatment (fixed at $11.09/treatment), processing ($20.00/animal), yardage ($0.20/animal/d), and feed costs ($0.88/kg of gain). We used the mean price for feeder calves during the study time period, which allowed us to account for the cattle cycle and to remove any variation in net return that would be caused by fluctuations in feeder calf prices.
Statistical analysis—Associations between the number of treatments after an initial diagnosis of BRD and demographic variables with the calculated net return values were analyzed via a generalized linear mixed model in a statistical software program.a The model included a Gaussian distribution, identity link function, maximum likelihood estimation, and Newton-Raphson with Ridging optimization technique. Random intercepts for cohort and year were included in the model to control for clustering of cattle nested within cohorts and cohorts nested within years.
On the basis of a causal diagram defined a priori, the number of treatments after an initial diagnosis of BRD (modeled as a polychotomous variable that included the categories 0, 1, 2, and ≥ 3) was considered the main predictor of interest. Variables pertaining to season (winter, January through March; spring, April through June; summer, July through September; and fall, October through December), sex (heifers and steers), and arrival weight class (categorized on the basis of weight ranges used for prices of US feeder cattle [136 to 180 kg, 181 to 226 kg, 227 to 272 kg, 273 to 318 kg, 319 to 363 kg, 364 to 408 kg, and > 409 kg]) were considered to be exposure-independent variables as well as potential a priori confounders; as such, they were added to the main effects model.12 The number of treatments after an initial diagnosis of BRD and arrival weight, which were originally recorded on a continuous scale, were categorized to avoid departures from the linearity assumption. Number of DOF, ADG, and carcass characteristics (ie, HCW, rib eye muscle area, back fat, CYG, and kidney, pelvic, and heart fat) were considered intervening variables in causal models for net returns because they could intervene in the causal pathway between arrival weight class and net return outcome. Thus, they were not included in the final model for net returns because their inclusion would have prevented us from correctly estimating the actual causal effects of arrival weight class on the outcome.12
During model building, an initial main effects model was built by including all predictors that were significantly associated with the outcome via univariable screening (P ≤ 0.10). A pairwise correlation analysis via Pearson and Spearman rank correlation statistics was performed among all variables significantly associated with the outcome for the univariable screening. When the value of the correlation statistic between 2 presumably independent variables was 0.8 or greater, only one of the variables was selected for inclusion in the multivariable model on the basis of biological plausibility and completeness of data.11 A manual backward-elimination procedure was then conducted until only significant (P ≤ 0.05; Wald χ2 test) main effects and confounding variables remained. All possible 2-way interaction terms between the main predictor of interest (number of treatments after an initial diagnosis of BRD) and confounders established a priori (ie, sex, season of arrival, and arrival weight class) were tested for significance (P < 0.05) via a manual forward-selection procedure.
Normal quantile plots of best linear unbiased predictors for the random variables in the model (ie, cohort and year of arrival) and residual plots of individual-level residuals were visually examined to assess general model fit and to identify potential outliers and influential observations. Coefficients and 95% CIs for those coefficients were calculated.
Models for performance and carcass trait outcomes—Performance outcomes included ADG, HCW, CYG, and QG. The ADG was calculated via the following equation: ADG = (total live weight at slaughter − arrival weight)/number of DOF). The HCW corresponded to the unchilled carcass weight measured shortly after slaughter. Both variables, which were originally recorded in pounds, were converted to kilograms and modeled on a continuous scale. The CYG was determined by use of an equation proposed by the USDA and described elsewhere9,13; the CYG was also modeled on a continuous scale. Quality grades were defined in accordance with USDA QG standards for beef.14 There were 2 categories for QG: choice or better (ie, prime and choice grades) and less than choice (ie, select grade).
To determine whether differences in a performance variable (ADG) or carcass traits (HCW, CYG, and QG) could be explained by the number of treatments during the feeding phase after an initial diagnosis of BRD, associations between the number of treatments and demographic variables determined at arrival to the feedlot with continuous outcomes (ie, ADG, HCW, and CYG) were modeled via a generalized linear mixed model with a Gaussian distribution, identity link, and maximum likelihood estimation in a statistical software program.a For the outcome pertaining to the probability of obtaining a carcass with a QG of choice or better, a generalized linear mixed model with a binary distribution, logit link, and maximum likelihood estimation was used. To account for the hierarchical structure of the data (ie, cattle nested within cohorts and cohorts nested within years), random intercepts for cohort and year of arrival were included in all models.
The number of treatments after an initial diagnosis of BRD was the main predictor of interest and determined on the basis of our causal models for performance and carcass characteristics. Number of DOF was not included in the final models because it was considered an intervening variable and could intervene in the causal pathway between arrival weight class and outcomes for performance variables and carcass traits. Sex, season of arrival, and arrival weight class were considered a priori confounders and were included in all main effects models. In addition, 2-way interaction terms between the main predictor of interest and a priori confounders were tested for significance (P < 0.05) via a manual forward-selection procedure. Model building and residual diagnostic analysis were conducted via the same steps described previously for the net returns model. Least squares means and 95% CIs for those estimates were computed for continuous outcomes, whereas ORs and their respective 95% CIs were estimated for predictors included in the final multivariable logistic regression model for the probability of obtaining a carcass with a QG of choice or better.
Results
Animals—Information for 212,867 cattle from 2,855 cohorts was used in the analysis. The mean ± SEM number of cattle per cohort was 75 ± 0.8 (median, 68 cattle; range, 1 to 349 cattle). There were 85,612 (40.2%) heifers and 127,255 (59.8%) steers. Descriptive statistics for net returns, performance traits, and carcass traits on the basis of the number of antimicrobial treatments after an initial diagnosis of BRD were calculated (Table 1).
Descriptive statistics of outcomes determined on the basis of the number of treatments after an initial diagnosis of BRD for cattle in a commercial feedlot.
| No. of treatments/animal* | Cattle (No. [%]) | Net return (US $; mean ± SD) | ADG (kg; mean ± SD) | HCW (kg; mean ± SD) | CYG (mean ± SD) | QG† | |
|---|---|---|---|---|---|---|---|
| Choice or better (No. [%]) | Less than choice (No. [%]) | ||||||
| 0 | 182,394 (85.68) | 30.08 ± 66.57 | 1.47 ± 0.28 | 340.89 ± 38.25 | 2.89 ± 0.74 | 84,968 (46.82) | 96,496 (53.18) |
| 1 | 21,741 (10.21) | 15.84 ± 67.12 | 1.36 ± 0.23 | 336.33 ± 40.13 | 2.79 ± 0.73 | 8,165 (37.67) | 13,510 (62.33) |
| 2 | 5,836 (2.74) | 2.11 ± 67.86 | 1.30 ± 0.21 | 335.17 ± 40.49 | 2.75 ± 0.72 | 1,918 (32.93) | 3,906 (67.07) |
| ≥ 3 | 2,896 (1.36) | −47.79 ± 82.23 | 1.24 ± 0.19 | 329.54 ± 42.02 | 2.62 ± 0.78 | 805 (27.92) | 2,078 (72.08) |
Cattle were never treated because of BRD or other diseases, or cattle were treated once, twice, or ≥ 3 times because of BRD during the first 100 days after arrival at the feedlot.
The denominator for QG is 211,846 cattle. Choice or better is the prime and choice grades, and less than choice is the select grade.
The cumulative proportion of cattle with BRD during the first 100 days after arrival at the feedlot was 30,473 of 212,867 (14.3%). Of those 30,473 affected cattle, 21,741 (71.3%) were treated only once, 5,836 (19.2%) were treated twice, and 2,896 (9.5%) were treated ≥ 3 times (Table 1). The majority (182,394/212,867 [85.7%]) of cattle were not treated because of BRD or any other disease during the first 100 days after arrival at the feedlot. For cattle that received antimicrobial treatments because of BRD, the mean ± SEM number of treatments was 1.6 ± 0.01 (range, 1 to 12 treatments).
Data from 8,316 records were removed from the dataset. This included records of 5,368 cattle for which there was an initial diagnosis of BRD but that did not receive antimicrobial treatment (thus they did not meet the definition of a BRD case) or for which there was an initial diagnosis of BRD but that did not receive antimicrobial treatment but then subsequently received antimicrobial treatment because of another illness; the removed data also included 2,250 cattle that had a disease other than BRD. Records were removed for 698 cattle for which there was an initial diagnosis of BRD and treatment but subsequent antimicrobial treatments for diseases other than BRD. The most frequently diagnosed diseases other than BRD were gastrointestinal tract conditions (eg, bloat), locomotor conditions (eg, lameness), and other problems (eg, abnormal behavioral activity in steers).
The mean ± SEM number of DOF throughout the entire feeding period for all cattle was 163 ± 0.1 (median, 157 DOF; range, 27 to 333 DOF). The mean number of DOF for cattle that were never treated or that were treated once, twice, or ≥ 3 times was 157.9 ± 0.1, 187.8 ± 0.3, 198.6 ± 0.5, and 203.8 ± 0.8, respectively. The mean number of days at time of first treatment after an initial diagnosis of BRD was 30.0 ± 0.14 days after arrival at the feedlot (median, 22 days after arrival).
Net returns model—Variables significantly associated with net returns in the final multivariable model were number of treatments after an initial diagnosis of BRD, sex, arrival weight class, season at time of arrival, and a 2-way interaction between season at time of arrival and the number of treatments after an initial diagnosis of BRD (Figure 1; Table 2). For all seasons, there were significantly higher net returns for cattle that never received treatments than for cattle that were treated ≥ 1 time after an initial diagnosis of BRD. However, the effect of the number of treatments after the initial diagnosis of BRD on net returns depended on the season of arrival at the feedlot, as indicated by a 2-way interaction. Overall, for all categories of the number of treatments after an initial diagnosis of BRD, net returns were higher for calves arriving at the feedlot during the fall and summer and lower for calves arriving at the feedlot during the spring and winter. There were significantly (P < 0.001) higher net returns ($32.97; 95%CI, $31.15 to $34.78) for heifers than for steers. In addition, net returns increased significantly (P < 0.001) as arrival weight increased.
Associations between demographic characteristics and number of treatments after an initial diagnosis of BRD with net returns, as determined via a multivariable final model.
| Variable | LSM | SEM | 95% CI |
|---|---|---|---|
| No. of treatments after an initial diagnosis of BRD* | |||
| 0 | 30.37 | 4.53 | 21.50 to 39.24 |
| 1 | 17.79 | 4.57 | 8.83 to 26.74 |
| 2 | 1.93 | 4.76 | −7.40 to 11.25 |
| ≥ 3 | −45.52 | 4.91 | −55.14 to −35.91 |
| Sex | |||
| Steer | −15.34 | 4.58 | −24.33 to −6.36 |
| Heifer | 17.62 | 4.60 | 8.61 to 26.63 |
| Arrival weight class (kg) | |||
| 136–180 | −22.37 | 4.75 | −31.68 to −13.06 |
| 181–226 | −18.43 | 4.60 | −27.44 to −9.42 |
| 227–272 | −6.63 | 4.57 | −15.59 to 2.34 |
| 273–318 | 3.30 | 4.57 | −5.65 to 12.26 |
| 319–363 | 12.08 | 4.57 | 3.12 to 21.04 |
| 364–408 | 15.82 | 4.58 | 6.84 to 24.81 |
| > 409 | 24.21 | 4.63 | 15.14 to 33.28 |
| Season† | |||
| Fall | 13.12 | 4.59 | 4.11 to 22.12 |
| Winter | −6.66 | 5.01 | −16.48 to 3.16 |
| Spring | −5.86 | 4.89 | −15.45 to 3.73 |
| Summer | 3.96 | 4.63 | −5.12 to 13.04 |
| No. of treatments after an initial diagnosis of BRD by season | |||
| Winter 0 | 23.75 | 4.64 | 14.66 to 32.84 |
| Winter 1 | 9.54 | 4.95 | −0.17 to 19.24 |
| Winter 2 | −7.75 | 6.50 | −20.49 to 4.99 |
| Winter ≥ 3 | −52.19 | 7.50 | −66.89 to −37.48 |
| Spring 0 | 26.48 | 4.70 | 17.27 to 35.69 |
| Spring 1 | 11.92 | 4.97 | 2.18 to 21.65 |
| Spring 2 | −7.47 | 5.69 | −18.63 to 3.68 |
| Spring ≥ 3 | −54.34 | 6.19 | −66.47 to −42.21 |
| Summer 0 | 31.83 | 4.59 | 22.83 to 40.83 |
| Summer 1 | 20.22 | 4.65 | 11.11 to 29.33 |
| Summer 2 | 6.37 | 4.81 | −3.06 to 15.80 |
| Summer ≥ 3 | −42.56 | 5.00 | −52.36 to −32.76 |
| Fall 0 | 39.41 | 4.57 | 30.45 to 48.37 |
| Fall 1 | 29.49 | 4.60 | 20.48 to 38.50 |
| Fall 2 | 16.56 | 4.69 | 7.37 to 25.75 |
| Fall ≥ 3 | −33.00 | 4.84 | −42.48 to −23.52 |
Net returns are US $/animal.
Cattle were never treated because of BRD or other diseases, or cattle were treated once, twice, or ≥ 3 times because of BRD during the first 100 days after arrival at the feedlot.
Seasons were defined as follows: winter, January through March; spring, April through June; summer, July through September; and fall, October through December.
LSM = Least squares mean.
Estimated least squares mean ± SEM of net returns by season of arrival and number of treatments after an initial diagnosis of BRD for cattle at a commercial feedlot. Cattle were never treated because of BRD or other diseases (diamonds and solid line) or were treated once (squares and dotted line), twice (triangles and dashed line), or ≥ 3 times (squares and dotted-and-dashed line) because of BRD during the first 100 days after arrival at the feedlot. Seasons were defined as follows: winter, January through March; spring, April through June; summer, July through September; and fall, October through December.
Citation: American Journal of Veterinary Research 74, 2; 10.2460/ajvr.74.2.300
Models for outcomes of performance variables and carcass traits—Associations between the number of treatments because of BRD and demographic factors with outcomes for performance variables and carcass traits were assessed via independent multivariable mixed-effects models to investigate whether differences in performance and carcass traits were associated with the number of treatments after an initial diagnosis of BRD. The number of treatments after an initial diagnosis of BRD, sex, season at time of arrival, weight class at time of arrival, and a 2-way interaction between season at time of arrival and number of treatments because of BRD were significantly associated with ADG (Table 3). The effect of the number of treatments after an initial diagnosis of BRD on ADG depended on the season of arrival at the feedlot. For all seasons, cattle that were never treated had a significantly higher ADG than did cattle treated ≥ 1 time with an antimicrobial after an initial diagnosis of BRD. Overall, for cattle that were never treated or that received 1 antimicrobial treatment because of BRD, ADG was significantly higher for cattle arriving during spring and winter, compared with the ADG for cattle arriving during summer and fall. Cattle receiving 2 treatments had a higher ADG when they arrived at the feedlot in the spring, but ADG remained almost unchanged for cattle receiving 2 treatments that arrived at the feedlot during the fall, winter, and summer (Figure 2). Heifers had a significantly lower (0.13 kg lower) ADG than did their male counterparts. The ADG increased significantly as the arrival weight increased, compared with the ADG of cattle in the lightest weight class (ie, 136 to 180 kg).
Association between number of treatments after an initial diagnosis of BRD, season of arrival at the feedlot, sex, and arrival weight class with economic, performance, and carcass outcomes for cattle at a commercial feedlot.
| Variable | Net return (US $; LSM ± SEM) | ADG (kg; LSM ± SEM) | HCW (kg; LSM ± SEM) | CYG (LSM ± SEM) | QG‡ (OR [SE]) |
|---|---|---|---|---|---|
| No. of treatments after an initial diagnosis of BRD* | |||||
| 0 | 30.37 ± 4.53 | 1.46 ± 0.005 | 335.85 ± 3.51 | 2.87 ± 0.04 | Ref |
| 1 | 17.79 ± 4.57 | 1.40 ± 0.006 | 335.38 ± 3.53 | 2.81 ± 0.04 | 0.86 (0.98) |
| 2 | 1.93 ± 4.76 | 1.34 ± 0.007 | 333.61 ± 3.65 | 2.77 ± 0.05 | 0.72 (0.97) |
| ≥ 3 | −45.52 ± 4.91 | 1.29 ± 0.008 | 323.14 ± 3.71 | 2.67 ± 0.05 | 0.56 (0.96) |
| Season† | |||||
| Fall | 13.12 ± 4.59 | 1.35 ± 0.006 | 342.29 ± 3.56 | 2.86 ± 0.05 | Ref |
| Winter | −6.66 ± 5.01 | 1.39 ± 0.010 | 326.43 ± 3.69 | 2.72 ± 0.05 | 1.17 (0.96) |
| Spring | −5.86 ± 4.89 | 1.42 ± 0.010 | 326.97 ± 3.68 | 2.70 ± 0.05 | 0.98 (0.96) |
| Summer | 3.96 ± 4.63 | 1.34 ± 0.007 | 332.29 ± 3.58 | 2.84 ± 0.05 | 1.10 (0.97) |
| Sex | |||||
| Steer | −15.34 ± 4.58 | 1.44 ± 0.006 | 343.01 ± 3.54 | 2.75 ± 0.05 | Ref |
| Heifer | 17.62 ± 4.60 | 1.31 ± 0.006 | 320.98 ± 3.56 | 2.82 ± 0.05 | 1.93 (0.97) |
| Arrival weight class (kg) | |||||
| 136–180 | −22.37 ± 4.75 | 1.32 ± 0.007 | 282.16 ± 3.67 | 2.53 ± 0.05 | Ref |
| 181–226 | −18.43 ± 4.60 | 1.33 ± 0.006 | 300.01 ± 3.55 | 2.61 ± 0.05 | 0.88 (0.95) |
| 227–272 | −6.63 ± 4.57 | 1.35 ± 0.006 | 315.77 ± 3.53 | 2.68 ± 0.05 | 0.84 (0.95) |
| 273–318 | 3.30 ± 4.57 | 1.37 ± 0.006 | 332.67 ± 3.53 | 2.76 ± 0.05 | 0.86 (0.95) |
| 319–363 | 12.08 ± 4.57 | 1.40 ± 0.006 | 349.10 ± 3.55 | 2.86 ± 0.05 | 0.93 (0.95) |
| 364–408 | 15.82 ± 4.58 | 1.42 ± 0.006 | 361.48 ± 3.63 | 2.96 ± 0.05 | 0.98 (0.95) |
| > 409 | 24.21 ± 4.63 | 1.42 ± 0.006 | 382.78 ± 4.00 | 3.05 ± 0.05 | 0.99 (0.94) |
| No. of treatments -by-season interaction§ | Yes | Yes | Yes | Yes | No |
| No. of treatments-by-sex interaction§ | No | No | Yes | No | No |
| No. of treatments-by-arrival weight class interaction§ | No | No | Yes | No | No |
Probability that the QG is choice or better versus select.
Analysis revealed that there was (yes) or was not (no) a significant (P < 0.05) interaction between the 2 factors.
Ref = Referent category.
See Tables 1 and 2 for remainder of key.
Estimated least squares mean ± SEM ADG by season of arrival and number of treatments after an initial diagnosis of BRD for cattle at a commercial feedlot. See Figure 1 for key.
Citation: American Journal of Veterinary Research 74, 2; 10.2460/ajvr.74.2.300
Variables significantly associated with HCW included the number of treatments after an initial diagnosis of BRD, sex, season of arrival, weight class, and 2-way interactions between number of treatments and sex, season of arrival, and weight class (Figure 3; Table 3). Overall, HCW was higher for cattle arriving during the fall than for cattle arriving during the winter, spring, and summer, regardless of the number of treatments they received because of BRD. Although there were no significant differences in HCW among seasons for cattle that were never treated or that received 1 or 2 treatments, each of these groups had a significantly higher HCW than did cattle that received ≥ 3 treatments because of BRD. Regardless of the number of treatments after an initial diagnosis of BRD, heifers had a significantly lower HCW, compared with the HCW for steers. Within each sex, HCW was not significantly different among cattle that were never treated or that received 1 or 2 treatments because of BRD, but each of these groups had a significantly higher HCW, compared with the HCW for cattle treated ≥ 3 times because of BRD. There was a linear increase in HCW as arrival weight increased, but the effect of weight at the time of arrival depended on the number of treatments after an initial diagnosis of BRD. Although HCW did not differ significantly between cattle that were never treated or that received 1 treatment because of BRD, each of those groups had a significantly higher HCW than for cattle receiving ≥ 3 treatments because of BRD. Cattle receiving 2 or ≥ 3 treatments had a departure from the linear association for HCW in those that weighed more at the time of arrival (categories 364 to 408 kg and > 409 kg).
Estimated least squares mean ± SEM HCW by season of arrival (A), sex (B), and weight class (C) and number of treatments after an initial diagnosis of BRD for cattle at a commercial feedlot. Notice that the scale on the y-axis differs among panels. See Figure 1 for remainder of key.
Citation: American Journal of Veterinary Research 74, 2; 10.2460/ajvr.74.2.300
For CYG, the number of treatments after an initial diagnosis of BRD, sex, season at time of arrival, arrival weight class, and a 2-way interaction between number of treatments after an initial diagnosis of BRD and season of arrival were significant (Table 3). Overall, CYG was higher for cattle that were never treated, compared with the CYG for cattle receiving ≥ 1 treatment because of BRD, across all arrival seasons (Figure 4). However, the effect of the number of treatments after an initial diagnosis of BRD on CYG depended on the season of arrival. Cattle arriving at the feedlot during the fall and summer had a significantly higher CYG than did cattle arriving during the winter and spring, regardless of the number of treatments because of BRD. Although there were no significant differences in CYG for cattle that were never treated or that received 1 or 2 treatments within each arrival season, the CYGs for those 3 groups were significantly higher than the CYG for cattle receiving ≥ 3 treatments, except for cattle that arrived during winter. The CYG for heifers was significantly higher (0.07 units higher) than the CYG for steers. The CYG increased significantly as arrival weight increased.
Estimated least squares mean ± SEM CYG by season of arrival and number of treatments after an initial diagnosis of BRD for cattle at a commercial feedlot. See Figure 1 for key.
Citation: American Journal of Veterinary Research 74, 2; 10.2460/ajvr.74.2.300
The probability of obtaining a QG of choice or better was significantly associated with the number of treatments after an initial diagnosis of BRD, sex, season of arrival at the feedlot, and arrival weight class (Table 3). No significant interactions were detected. The odds of obtaining a QG of choice or better were significantly (P < 0.001) lower for cattle that received ≥ 1 treatment because of BRD (OR, 0.86, 0.72, and 0.56 for 1, 2, and 3 treatments respectively) than for cattle that were never treated. Heifers were significantly more likely (OR, 1.93) than steers to have a carcass with a QG of choice or better. Cattle with a lower weight at arrival (136 to 180 kg) were more likely to have a carcass with a QG of choice or better, compared with the likelihood for cattle with an arrival weight of 181 to 226 kg (OR, 0.88; P = 0.007), 227 to 272 kg (OR, 0.84; P < 0.001), and 273 to 318 kg (OR, 0.86; P = 0.002). Odds of obtaining a QG of choice or better were significantly higher for cattle arriving at the feedlot during the winter (OR, 1.17; P < 0.001) and summer (OR, 1.10; P = 0.004), compared with the odds for cattle arriving during the fall.
Discussion
In the present study, novel information was found on the association between the number of treatments after an initial diagnosis of BRD and outcomes of economic importance. Smaller-scale studies6,15,16 have addressed the impact of treatments on performance between healthy and sick cattle; however, those investigators did not evaluate the number of treatments because of BRD via a large dataset over multiple years and classes of cattle or in a multivariable framework. Multivariable assessment provides a more comprehensive mechanism to assess economic consequences of BRD for heterogeneous populations of cattle by identifying demographic risk factors as well as potential confounders and effect modifiers.
Analysis of the results of the present study indicated that the number of treatments after an initial diagnosis of BRD was associated with net returns, and this effect was modified by season of arrival when evaluated in a model that included sex and arrival weight class. Previous studies6,15 were restricted to only steers or only heifers; thus, the effect of sex could not be evaluated. In the study reported here, we found that after accounting for arrival weight class, season of arrival, and the number of treatments after an initial diagnosis of BRD, there were higher net returns for heifers than for steers. In addition, net returns significantly increased as arrival weight increased. Arrival weight is considered a proxy for age and an important factor for predicting subsequent health risks.17,18 Cattle that are young (and likely with a lower weight) at the time of arrival may be more susceptible to stress and more prone to become sick than are cattle with a higher body weight,19,20 which may lead to increased costs for treatments and lower profits in terms of weight gain and performance traits.
Similar to results of other studies,7,21 net returns in the present study significantly decreased as the number of treatments because of BRD increased, compared with the net returns for cattle that were never treated; however, the effect of this predictor on net returns in the study reported here depended on the season of arrival. Within each category for the number of treatments after an initial diagnosis of BRD, net returns were higher for cattle arriving during the fall and summer than for cattle arriving during the spring and winter. Risk of developing BRD is reportedly higher in the summer and fall,22,23 compared with the risk for cattle arriving in the winter and spring. Unexpectedly, net returns were higher for cattle arriving in seasons with a higher BRD incidence (ie, fall and summer). Perhaps the reason for the inverse relationship between season of arrival (and BRD risk) and net returns relates to misclassification of disease (ie, the effects of incorrectly classifying the outcome or disease status)12 when risk of BRD increases. Season can be a proxy for weather, and weather conditions can affect BRD morbidity rates24 and subsequent feed intake, weight gain, and carcass characteristics, which may explain the lower net returns during the winter. Similarly, higher mortality ratios associated with respiratory tract disorders were observed in the fall months in another study.25 It is possible that higher death losses incurred during the fall and summer could have been factored into the purchase price for calves, compared with the purchase price paid for calves in the winter or spring; however, this would have increased the economic returns on the basis of lower feeder prices for calves purchased in the fall and summer.
Net returns were calculated on the basis of costs derived from 1996 to 2006. These dates were selected to provide a wide range of cattle and feed prices, and mean values were used to calculate economic returns; therefore, although the estimated net returns would differ for various market conditions, the relationships between categories (ie, number of treatments because of BRD) should have remained similar.
Net returns can be affected by differential use of metaphylactic treatments and death losses across seasons, sex, arrival weight, and years.26 The metaphylactic use of antimicrobials may reduce morbidity risk and increase performance; however, the expected reduction in disease risk, and subsequent treatments, depends on the product administered, the morbidity risk, and the demographic characteristics of the cattle receiving the intervention.26 Unfortunately, information on use of metaphylactic treatments was not available in the database of the present study. Although the number of deaths affects net returns, cattle deaths were not included in the economic model because performance and carcass data were not available for cattle that died. We recognize that the estimated effects of treatments on net returns (especially for cattle receiving 2 or 3 treatments because these cattle would be more likely to die of respiratory tract disease) may be conservative and potentially overestimated. Similarly, expected death losses could have affected feeder prices during specific seasons, which consequently would have affected net returns.
Similar to results in previous studies,8,9 net returns in the present study were influenced by the number of treatments because of BRD, HCW, and ADG. However, we included arrival factors that drive the economic differences between sick and healthy cattle. Although previous studies6,7 indicated that the timing of treatment was associated with net returns, time of treatment administration was not evaluated in the present study because it was considered an intervening variable in the pathway between arrival weight and net returns. Raw data on net returns for the study reported here are similar to those reported in other studies6,7; however, when net returns were assessed via a multivariable framework that accounted for other demographic predictors, the effect of the number of antimicrobial treatments after an initial diagnosis of BRD on net returns depended on the season cattle arrived at the feedlot. The smaller differences in net returns between cattle that were never treated or that were treated once, compared with net returns for cattle treated 2 or ≥ 3 times, especially during periods of higher risk for developing BRD (ie, fall and summer), raised the concern of potential misclassification of disease status.27 In periods of high BRD risk (eg, fall), a decrease in the specificity of the assessment of illness, which would cause a higher rate of false-positive classifications (ie, cattle identified as sick when they were actually free of disease), could have been responsible for the small differences in net returns between cattle that were never treated and cattle that were treated once because some of the cattle receiving 1 treatment may have been free of disease. Similarly, cattle that had early or less pronounced clinical signs of disease could have been misclassified as not being sick (ie, were never treated); thus, the sensitivity of detection was also inferior during periods of high risk for developing BRD. As indicated in another study,27 the relatively low sensitivity and specificity of the assessment of clinical illness are expected, given the subjectivity of the method and intrinsic behavior characteristics of cattle. Such misclassifications were fewer for the cattle receiving 2 or ≥ 3 treatments because these categories could be considered a sequential interpretation of BRD treatment events.
Although there is no clear mechanism that explains how disease affects carcass traits,28 there is evidence that relapses of respiratory disease have a negative effect on performance and compensatory gain.7,15,16 However, it is not known whether performance is decreased because of the illness or whether cattle with poor performance are more likely to become sick.29
In other studies, nontreated cattle had a higher ADG than cattle treated once30,31 or more than once15,31 because of respiratory disease. It is rational that ADG decreases with an increase in the number of treatments because cattle re-treated as a result of reinfection or disease relapse would lose more weight and thus have a lower sale weight while requiring more time in the finishing phase, compared with outcomes for healthy cattle or cattle treated fewer times.15,28,29 Although the effect of the number of treatments after an initial diagnosis of BRD across seasons on ADG remained almost constant, cattle arriving at the feedlot during the spring and winter had higher ADGs than did cattle arriving in the fall and summer. Perhaps the effects of weather characteristics, feed, and water availability can explain these fluctuations in ADG. Interestingly, cattle arriving at the feedlot during the winter and spring had higher ADGs but yielded lower net returns. Perhaps producers pay more for cattle in the winter because it is expected they may have lower mortality rates. Cattle arriving during the winter eat greater quantities of feed but convert it less efficiently; hence, as these cattle gain weight, ADG will eventually decrease and the total cost for the total weight gained will increase, which explains the lower net returns.32
In previous studies,7,15,16 researchers found that cattle treated ≥ 1 time because of BRD had a lower HCW than did untreated cattle. However, in the present study, the effect of the number of treatments on HCW depended on sex, arrival weight class, and the season of arrival. Overall, cattle treated ≥ 3 times had a lower HCW in heifers with a heavier arrival weight (> 364 kg) that arrived at the feedlot during the winter and spring. Cattle receiving 0, 1, or 2 treatments may have been able to compensate, which was reflected in their carcass weight, compared with results for cattle receiving ≥ 3 treatments.
Analysis of the data indicated that CYG within each arrival season did not differ among cattle receiving 0, 1, or 2 treatments, but these 3 groups of cattle had significantly higher CYGs than did cattle receiving ≥ 3 treatments and arriving during the fall, spring, or summer, with cattle arriving during the fall having the highest CYG, regardless of the number of treatments. Similarly, other researchers7,15,16 found that CYG decreased as the number of treatments increased. Although better prices may be obtained as a carcass becomes leaner, this effect is likely diminished by a decrease in carcass weight.
The odds of obtaining a QG of choice or better were significantly higher in heifers than steers, cattle arriving during the winter and summer than cattle arriving during the fall, and cattle in the lightest weight class at arrival than in cattle in heavier weight classes at arrival. Similar to the findings in another study,15 in which cattle treated because of BRD had a higher percentage of carcasses with a standard grade than did nontreated cattle, analysis of data in the present study indicated that cattle receiving ≥ 1 treatment after an initial diagnosis of BRD had lower odds of having carcasses with a QG of choice or better than did cattle that were never treated.
The effect of the number of treatments after an initial diagnosis of BRD during the feeding period on net returns depended on the season of arrival, an effect that was also evident for the ADG, HCW, and CYG models. The differences in net returns among the treatment groups may be explained primarily by changes in HCW and QG. Given that these variables are embedded in the grid price component of the net returns equation, it appears reasonable that an increase in the values for HCW and QG will be reflected in an increase in net returns.
Yield grades and QGs are summary measures of carcass traits commonly used to evaluate health interventions. Yield grades can be converted to the nearest integer; thus, they could be analyzed via an ordinal regression. Given that CYG values (on a continuous scale) were available in the database and assessment of the residuals of the final model indicated that the normality assumption was met, linear regression was used to model associations between predictors and yield grade. Although QG is an ordinal outcome, we opted to dichotomize it as choice or better versus less than choice on the basis that this is a commonly used cutpoint in the industry as pertains to the grades sought in a carcass. In a study33 conducted to evaluate the use of various statistical approaches to analyze QG and yield grade outcomes, investigators concluded that even though most of the statistical methods applied result in similar point estimates, some procedures (eg, ordinal regression) offer more advantages when modeling these outcomes than does stratification, linear regression, or logistic regression. Nonetheless, we selected the statistical approach on the basis of the available data, objectives of the study, and ability to realistically interpret and apply the findings on the basis of cutoff values that are relevant to livestock production.
The present study was unique because it comprised a large amount of individual-animal data from a US commercial feedlot; these data are not always readily available in feedlot operational databases. Although animal data from a large commercial feedlot were used in the study, further research with data from multiple feedlots is warranted to improve the external validity of our findings.
The retrospective, cross-sectional nature of the analysis prevented us from drawing direct causal inferences. Furthermore, the directionality of the association between the number of treatments after an initial diagnosis of BRD with net returns and performance and carcass traits cannot be determined because treatment was not a time-invariant exposure (ie, it changed with time). Therefore, it is difficult to determine cause and effect because of the reverse-causation problem.12 In this case, and as hypothesized in another report,29 we cannot differentiate whether the increasing number of treatments after an initial diagnosis of BRD caused the decreased performance (ADG, HCW, CYG, and QG) or whether cattle with decreased performance were more prone to develop disease. Regardless of the actual direction of the associations, control of factors associated with performance or development of disease will have a beneficial impact for both feedlot production and animal health.
Analysis of the data indicated that differences in net returns depended on costs incurred because of treatments, but they also differed on the basis of specific demographic characteristics of cattle at the time of arrival at the feedlot. The methods used to calculate the costs related to net returns for each animal were based on cost of gain calculated by use of standard assumptions related to dry-matter intake; therefore, if dry-matter intake was altered by disease status, the net returns in diseased cattle could have been slightly overestimated. Similarly, performance and carcass traits were determined by the use of similar factors (specifically, carcass traits pertaining to weight and yield grade), which were likely responsible for the differences in net returns among cattle that received different numbers of treatments. It is important to mention that the carcass characteristics (ie, HCW, CYG, and QG) modeled in the present study are relevant to cattle marketed on a grid basis but may not be as relevant if cattle are priced on a live-weight basis. However, economic modeling in this manner provides an estimate of the broader biological impacts of BRD, and the results reported here reflected a pattern observed in recent years whereby there has been a greater emphasis on improving the quality and consistency of beef. Thus, grid pricing has become more common and appealing to producers as these methods target improvements in pricing accuracy and provide compensation to producers who market better-quality cattle.
Knowledge of arrival factors (eg, sex, body weight, and season) that could determine differences in net returns and traits of economic importance can help feedlot managers increase profitability when they make decisions. Moreover, information on net returns can become a useful tool for producers when making marketing decisions and determining a course of action on the basis of fluctuations in the market values of cattle and feed.
ABBREVIATIONS
| ADG | Average daily gain |
| BRD | Bovine respiratory disease |
| CI | Confidence interval |
| CYG | Calculated yield grade |
| DOF | Days on feed |
| HCW | Hot carcass weight |
| QG | Quality grade |
Proc GLIMMIX, SAS, version 9.2, SAS Institute Inc, Cary, NC.
References
1. Griffin D, Perino L, Wittum T. Feedlot respiratory disease: cost, value of preventives and intervention, in Proceedings. 27th Annu Conf Am Assoc Bovine Pract 1995;157–160.
2. USDA. Part III: health management and biosecurity in US feedlots, 1999. Fort Collins, Colo: UDSA, APHIS, Veterinary Services, Centers for Epidemiology and Animal Health, National Animal Health Monitoring System, 2000. Available at: www.aphis.usda.gov/animal_health/nahms/feedlot/downloads/feedlot99/Feed-lot99_dr_PartIII.pdf. Accessed Oct 15, 2011.
3. Griffin D. Economic impact associated with respiratory disease in beef cattle. Vet Clin North Am Food Anim Pract 1997; 13: 367–377.
4. White BJ, Anderson JD, Larson RL, et al. Review: the cow-calf operation retained ownership decision. Prof Anim Sci 2007; 23: 18–28.
5. White BJ, Anderson JD, McKinley WB, et al. Factor price disparity and retained ownership of feeder cattle: an application of feedlot and carcass performance data to farm-level decision making. J Agric Appl Econ 2007; 39: 87–101.
6. Brooks KR, Raper KC, Ward CE, et al. Economic effects of bovine respiratory disease on feedlot cattle during backgrounding and finishing phases. Prof Anim Sci 2011; 27: 195–203.
7. Schneider MJ, Tait RG, Busby WD, et al. An evaluation of bovine respiratory disease complex in feedlot cattle: impact on performance and carcass traits using treatment records and lung lesion scores. J Anim Sci 2009; 87: 1821–1827.
8. Thompson PN, Stone A, Schiltheiss WA. Use of treatment records and lung lesion scoring to estimate the effect of respiratory disease on growth during early and late finishing periods in South African feedlot cattle. J Anim Sci 2006; 84: 488–498.
9. Babcock AH, White BJ, Dritz SS, et al. Feedlot health and performance effects associated with the timing of respiratory disease treatment. J Anim Sci 2009; 87: 314–327.
10. Smith RA. North American cattle marketing and bovine respiratory disease (BRD). Anim Health Res Rev 2009; 10: 105–108.
11. Livestock Marketing Information Center website. Oklahoma City feeder & stocker cattle summary. Available at: www.lmic.info. Accessed Oct 13, 2011.
12. Dohoo I, Martin W, Stryhn H. Confounding: detection and control. In: Veterinary epidemiologic research. 2nd ed. Charlottetown, PE, Canada: AVC Inc, 2009;271–322.
13. Busby D, Loy D. Called versus calculated yield grades. ISU beef research report. AS Leaflet R1730. Ames, Iowa: Iowa State University, 2000.
14. USDA. Standards for grades of slaughter cattle and standards for grades of carcass beef. Washington, DC: US Government Printing Office, 1996.
15. Gardner BA, Dolezal HG, Bryant LK, et al. Health of finishing steers: effects on performance, carcass traits and meat tenderness. J Anim Sci 1999; 77: 3168–3175.
16. Roeber DL, Speer NC, Gentry JG, et al. Feeder cattle health management: effects on morbidity rates, feedlot performance, carcass characteristics and beef palatability. Prof Anim Sci 2001; 17: 39–44.
17. Lechtenberg KF, Smith RA, Stokka GL. Feedlot health and management. Vet Clin North Am Food Anim Pract 1998; 14: 177–197.
18. Babcock AH, Renter DG, White BJ, et al. Temporal distributions of respiratory disease events within cohorts of feedlot cattle and associations with cattle health and performance indices. Prev Vet Med 2010; 97: 198–219.
19. Eicher SD. Transportation of cattle in the dairy industry: current research and future directions. J Dairy Sci 2001; 84 (suppl): E19–E23.
20. Swanson JC, Morrow-Tesch J. Cattle transport: historical, research, and future perspectives. J Anim Sci 2001; 79 (suppl E): E102–E109.
21. Fulton RW, Cook BJ, Step DL, et al. Evaluation of health status of calves and the impact on feedlot performance: assessment of a retained ownership program for postweaning calves. Can J Vet Res 2002; 66: 173–180.
22. Ribble CS, Meek AH, Janzen ED, et al. Effect of time of year, weather, and the pattern of auction market sales on fatal fibrinous pneumonia (shipping fever) in calves in a large feedlot in Alberta (1985–1988). Can J Vet Res 1995; 59: 167–172.
23. Ribble CS, Meek AH, Shewen PE, et al. Effect of transportation on fatal fibrinous pneumonia and shrinkage in calves arriving at a large feedlot. J Am Vet Med Assoc 1995; 207: 612–615.
24. Cernicchiaro N, White BJ, Renter DG, et al. Associations between distance traveled and feedlot performance, mortality, and risk of respiratory disease. J Anim Sci 2012; 90: 1929–1939.
25. Loneragan GH, Dargatz DA, Morley PS, et al. Trends in mortality ratios among cattle in US feedlots. J Am Vet Med Assoc 2001; 219: 1122–1127.
26. Nickell JS, White BJ. Metaphylactic antimicrobial therapy for bovine respiratory disease in stocker and feedlot cattle. Vet Clin North Am Food Anim Pract 2010; 26: 285–301.
27. White BJ, Renter DG. Bayesian estimation of the performance of using clinical observations and harvest lung lesions for diagnosing bovine respiratory disease in post-weaned beef calves. J Vet Diagn Invest 2009; 21: 446–453.
28. Larson RL. Effect of cattle disease on carcass traits. J Anim Sci 2005; 83 (suppl E): E37–E43.
29. Smith RA. Impact of disease on feedlot performance: a review. J Anim Sci 1998; 76: 272–274.
30. Bateman KG, Martin SW, Shewen PE, et al. An evaluation of antimicrobial therapy for undifferentiated bovine respiratory disease. Can Vet J 1990; 31: 689–696.
31. Morck DW, Merrill JK, Thorlakson BE, et al. Prophylactic efficacy of tilmicosin for bovine respiratory tract disease. J Am Vet Med Assoc 1993; 202: 273–277.
32. Tatum JD, Platter WJ, Bargen JL, et al. Carcass-based measures of cattle performance and feeding probability. Prof Anim Sci 2012; 28: 173–183.
33. Osterstock JB, MacDonald JC, Boggess MM, et al. Technical note: analysis of ordinal outcomes from carcass data in beef cattle research. J Anim Sci 2010; 88: 3384–3389.
