Economic analysis of randomized controlled trial data: a framework and feedlot cattle case study

Abstract

Livestock industry stakeholders rely on research, often randomized controlled trials, to make evidence-based decisions. Economic implications of interventions are often a major deciding factor for adoption by producers. However, economic analyses in beef feedlot trials are infrequently conducted and often suffer from inconsistent methodologies. Gaps in planning, execution, and reporting of economic assessments underscore the need for guidance and standardized approaches in conducting economic evaluations on RCT data. Our objective was to compare and contrast methodologies for assessing costs and benefits associated with livestock health and production trials and to provide scientific guidance, rationale, and recommendations for future conduct of economic analyses on RCT data. Several types of economic analyses are frequently used by agricultural economists, including cash flow budgets, enterprise budgets, gross margin analyses, cost-benefit and -effectiveness analyses, and partial budgets. Partial budgeting emerges as the most pragmatic strategy for RCT data, focusing on the marginal impact of alternative interventions or management strategies, aligning well with RCT objectives. We provided an example of applying a partial budget to an RCT conducted at a commercial beef feedlot using published data. All observed data for relevant animal performance, health, and carcass variables were included, regardless of their original statistical significance. The budget was applied to each experimental unit (pen), with net return as the final outcome, and analyzed statistically using linear mixed models. While simple partial budgets use fixed prices that may not represent economic risk, incorporating statistical analyses at the pen-level accounts for biological variability and error in the estimates. When warranted, other strategies to account for economic risk (e.g., sensitivity analysis, stochastic simulation) can be incorporated within a partial budget framework. To encourage robust and transparent reporting, future research should explicitly state the type of economic assessment, the values and sources of all prices and the timeframe they represent, the methodology used, and how analyses were conducted. By adopting more consistent and transparent economic evaluation methods, researchers can enhance the applicability of RCT findings, ultimately supporting stakeholders in making economically sound decisions.

By demonstrating how researchers can effectively evaluate the economic impacts of interventions in livestock from randomized controlled trials, this publication provides essential guidance to enhance research transparency and support stakeholders in making informed, economically sound decisions.

Introduction

Industry leaders in production livestock continually strive for advancements to improve operational efficiencies and sustainability. Sustainability can be broadly defined in 3 categories, which are: social, environmental, and economic sustainability (Goodland and Daly, 1996). Trade-offs exist among these categories, and stakeholders rely on research in order to make evidence-based decisions. When considering research outcomes in the animal sciences, economics is one of the most critical components, which fundamentally intersects with health and production outcomes (Dewsbury et al., 2022). In livestock production systems, economic considerations are paramount under an outcomes research framework, as they directly influence the adoption of interventions and management strategies by producers. The economic viability of a health or production intervention often serves as the deciding factor for its implementation by producers. In other words, no matter how innovative a new product or strategy may be, without economic incentive, adoption by producers may be limiting. Stakeholders must weigh the costs of adopting new practices against the potential benefits in terms of improved animal health, performance, and overall economic sustainability.

Despite the critical importance of economics in livestock decision-making, there is a notable inconsistency in the conduct and reporting of economic evaluations in animal health and production research. Saatkamp et al. (2016) highlighted the lack of a standardized framework for categorizing the costs associated with highly contagious livestock diseases, leading to difficulties in comparing studies and potential misinterpretations. They emphasized that without consistent methodologies rooted in sound economic principles, economic analyses can become fragmented and less useful for stakeholders. Further reinforcing this concern, Dixon et al. (2022) conducted a scoping review of experimental feedlot trials and found substantial variability and inconsistencies in the methods and reporting of economic assessments. In their review, which included only articles with some form of economic assessment, just 28% of publications stated an economic outcome as a primary or secondary outcome of interest. The type of economic assessment and methodology [e.g., sourcing of prices, calculation of values, and at what level(s)] was reported in 26% of trials; 48% reported methodology only; 19% only stated the type of economic assessment used; and 8% reported neither (Dixon et al., 2022). Selective use of data was apparent, as just 42% used all measured outcomes in the economic assessments; some used statistical tests for this selection, as 22% of articles included only statistically significant outcomes as variables.

Overall findings from Dixon et al. (2022) demonstrated that many studies lacked explicit statements of primary economic outcomes, did not adequately describe the methodologies used, and often failed to provide sufficient details for replication or critical evaluation. These are basic components needed for the assessment of reproducibility and validity by researchers (Sargeant et al., 2010). This inconsistency hampers the ability of researchers and industry stakeholders to fully understand and apply the findings of economic assessments, ultimately limiting their impact on industry practices and policy-making.

Randomized controlled trials (RCTs) have long been considered the gold standard for the evaluation of research outcomes from interventions, and rank near the top of the hierarchy of evidence for causal inference, just below systematic reviews and meta-analysis of multiple RCTs (Sargeant et al., 2014). Stakeholders commonly use findings from RCTs in livestock production to make better-informed decisions when applying interventions or implementing new management strategies. Therefore, integrating robust economic analyses into RCTs is essential for providing stakeholders with comprehensive information. Given the critical role that economic outcomes play in influencing producer decisions and the observed inconsistencies in current research practices, there is a clear need for more guidance and standardization in conducting economic analyses on RCT data. Establishing consistent methodologies and reporting standards will enhance the reliability and comparability of economic assessments, aiding stakeholders in making better-informed decisions based on robust and transparent economic evidence. Our objective was to compare and contrast methodologies for assessing costs and benefits associated with livestock health and production trials, and to provide scientific guidance, rationale, and recommendations for future conduct of economic analyses on RCT data.

Review of analysis strategies for economic assessments

There are several economic analysis methods commonly used in agricultural economics, each suited to particular goals. Scopes and potential relevance of these strategies to RCT data are briefly described here. Cash flow budgets focus on the timing of all incoming and outgoing funds over a defined period (e.g., a year), providing insights into an operation’s liquidity and ability to meet financial obligations as they occur (Madsen, 1983; Hardie et al., 1984; Trejo-Pech et al., 2022). Whole farm budgets, on the other hand, take a holistic approach by evaluating profitability across all primary and secondary enterprises, including livestock, crops, and other revenue streams, enabling assessment of resource allocation decisions or farm diversification (Rushton, 2007; Hansen and Nærland, 2017). Similarly, gross margin analysis evaluates revenue minus variable costs for different enterprises within an operation, except it excludes fixed or overhead costs that are shared across enterprises (Rushton, 2007). This simplification makes it somewhat more practical than whole-farm budgeting.

Enterprise budgets narrow the focus further, evaluating profitability within a single enterprise, such as a cow–calf or feedlot operation, by clearly distinguishing fixed or overhead costs (e.g., depreciation, labor) from variable costs (e.g., feed, veterinary expenses) (Madsen and Haider, 1983; Rushton, 2007). Because enterprise budgets focus on a single entity within a larger production framework, they may be of greater utility than, e.g., whole farm budgets, due to their simplicity and reliance on fewer assumptions that may change during a production cycle (generally 1 yr). Although valuable for strategic planning, the budgets described thus far often incorporate broad assumptions about costs and revenue sources that may not clearly isolate incremental financial effects directly attributable to a single management change or intervention.

The ultimate goal of a RCT is to measure the effect, or marginal change that occurs when applying alternative management strategies or interventions. Therefore, the structure and purpose of subsequent economic assessments should align with this goal. Given the limitations of previously described budgets, partial budgeting often emerges as the most pragmatic method for assessing marginal economic impacts in RCTs (Corbin and Griffin, 2006; Rushton, 2007).

Partial budgeting explicitly considers the incremental changes directly associated with implementing a new intervention (e.g., a change in management or production strategy, feed additive, medication, or precision technology), capturing 4 key categories: additional costs incurred, reduced costs, additional revenue generated, and reduced revenue resulting from the intervention (Rushton, 2007). The objective is not to assess profits or losses of an enterprise as a whole but to estimate the marginal net return difference (partial revenues minus partial costs) or change resulting from the alternative intervention(s). Partial revenues may represent total revenue in a partial budget, while partial costs likely will not represent total cost, as costs that are constant across interventions (e.g., property taxes) are excluded from the budget. The net economic impact is calculated by comparing the sum of additional costs and reduced revenue to the sum of additional revenue and reduced costs, and economic support for the proposed change is achieved when the net benefits exceed net costs. This streamlined focus aligns with the experimental design of RCTs, where researchers often seek to isolate treatment-specific effects on outcomes. An important distinction in partial budgeting is that, because we are only interested in the marginal impact of a change, only items that are suspected to be affected by the change need to be included in the budget (Corbin and Griffin, 2006). Inputs not expected to be modified by the change (e.g., most fixed costs) can be excluded. However, if there is uncertainty about whether an item is affected, it is prudent to include it in the budget.

Cost-benefit and cost-effectiveness analyses are additional approaches often used in agricultural and health economics, generally for more complex policy-related decisions (Rushton, 2007). Cost-benefit analysis compares the total monetized costs of a project or intervention against its total expected benefits (including nonmonetary outcomes converted to monetary terms) to determine overall economic viability (Harrison, 1996; Adamson et al., 2020). Alternatively, cost-effectiveness analysis compares costs directly with outcomes measured in natural units, such as weight gain or disease incidence reductions, making it useful for interventions that aim for similar endpoints but differ in their cost structures (Babo Martins and Rushton, 2014). While these methodologies can be valuable, both typically involve complexities that exceed the practical need for a straightforward evaluation of interventions within RCTs focused primarily on animal production outcomes.

A general limitation of the methods discussed thus far is their confinement to fixed production cycles or time intervals, potentially missing the inherent variability and uncertainty (i.e., risk) associated with fluctuating economic conditions or variable animal performance (Rushton, 2007). Advanced strategies have been developed to explicitly incorporate risk and uncertainty, including sensitivity analysis, break-even analysis, decision analysis, and stochastic simulation modeling. Sensitivity analysis assesses the robustness of conclusions by systematically varying key economic assumptions or input parameters, such as prices, costs, or animal performance variables, and examining the resulting effects on economic outcomes (Frey and Patil, 2002). Sensitivity analyses can be readily incorporated within a partial budget framework; for instance, Horton et al. (2024) used sensitivity analyses and RCT data to evaluate how different price scenarios influenced net returns of feedlot heifers fed for varying days-on-feed (DOF) and under different implant strategies. Break-even analysis, a form of sensitivity analysis, identifies the conditions under which the net benefit of an intervention equals zero, providing clarity on when an intervention becomes economically viable (Berry, 1972; Rushton, 2007).

Decision analysis, another advanced method, employs decision trees or similar models to explicitly represent and evaluate multiple outcomes resulting from different choices under uncertainty, integrating probabilities for various scenarios (Marsh, 1999; Morris, 1999). Dynamic, stochastic simulation modeling goes even further by assigning probability distributions to input variables and may account for co-dependencies between variables; repeated iterations are conducted, thus providing a comprehensive distribution of potential economic outcomes under diverse conditions (Corbin and Griffin, 2006). Such models have been employed in livestock economics to address complex, but pragmatic questions in the livestock production sector (Belasco et al., 2009; Dennis et al., 2018, 2020). While these models offer additional robustness, they are generally more complex and resource-intensive, and often not solely based on RCT data. These advanced methods that address risk are beyond the scope of this publication, as they are better suited for analyses where the primary objective is economic evaluation under uncertainty. In most RCTs, the primary objective is related to some aspect of animal health and production, with economics as a secondary or supplementary objective. Thus, simpler budgeting methods like partial budgeting typically suffice for most livestock RCTs.

To summarize, while multiple economic analysis methods are available, partial budgeting emerges as the most practical tool for evaluating the marginal economic impacts of interventions tested in RCTs focused on livestock health and production. Its focus on changes directly related to the intervention, combined with its simplicity and adaptability, makes it a method well aligned with RCT objectives. These features allow transparent reporting and enable improved research reproducibility. Furthermore, partial budgeting can be adapted or expanded using advanced risk-assessment methods, including sensitivity or stochastic simulation analyses, if warranted by research questions and economic context (Berry, 1972; Marsh, 1999; Morris, 1999; Corbin and Griffin, 2006; Rushton, 2007; Horton et al., 2024). Therefore, to illustrate the practicality and adaptability of partial budgeting, a case study using RCT data was detailed and executed.

Case study—applying a partial budget to RCT data

To provide an example and guidance on applying partial budgeting to evaluate economic implications from RCT data, we utilized data from Nickell et al. (2021) for the Oklahoma (OK) trial site. In brief, the objective of this trial was to evaluate the effects on feedlot steer health, performance, and carcass characteristics of bovine respiratory disease (BRD) control programs targeted at the animal level using predictive algorithms, compared to industry-standard pen-level management strategies (metaphylaxis or none) of BRD. The control programs were based on data from a precision agriculture technology (TECH) designed to predict animal-level risk of BRD using machine learning algorithms that process chute-side diagnostics (e.g., weight, heart and lung sounds, temperature) as input variables. Based on the algorithm, the TECH indicates whether each animal should be administered an antimicrobial at initial processing, and the system’s “aggressiveness” in indicating animals for BRD treatment can be adjusted, where a more aggressive threshold targets fewer animals to receive an antimicrobial (i.e., more aggressive equals more animals selected to NOT receive metaphylaxis that otherwise would). Thus, the proportion of animals within a pen that receives an antimicrobial based on the TECH will vary depending on the overall BRD risk of the pen and the programmed threshold.

The trial was conducted as a randomized complete block design with a 1-way factorial arrangement of 4 experimental treatments. Treatments consisted of negative control (NEG; no metaphylaxis), a positive control (META; metaphylaxis to all animals), TECH-high proportion (metaphylaxis administered to individual animals based on a conservative algorithm threshold), and TECH-low proportion (metaphylaxis administered to individual animals based on an aggressive algorithm threshold). A total of 1,980 beef steer calves were allocated across 28 pens, with 69 to 72 animals per pen. Calves were blocked by arrival, randomly allocated to pens within blocks, and pens were randomly assigned to treatments. The primary health outcomes were BRD morbidity and mortality. Different antimicrobials were used for first, second, and third clinical BRD treatments, but did not differ among experimental treatments (Table 1). Steers requiring more than 3 treatments for clinical BRD were removed from the trial (culled). Performance, health, and carcass outcomes were monitored through harvest. For additional information on the RCT used for our case study, readers are referred to Nickell et al. (2021).

Table 1.

Prices and their sources for use in a partial budget that was applied to feedlot steer pens from randomized controlled trial data

Item1	Price	Source
Cost
Feeder steer price, $/cwt	$163.70	LMIC, USDA-AMS data
Initial processing2
Chute charge, $/n	$1.50	Assumed
Ear tag, $/n	$0.39	Valleyvet.com
Vista 5 SQ, $/n	$1.17	Valleyvet.com
Vision 7 with Spur, $/n	$0.65	Valleyvet.com
Revalor-XS, $/n	$8.44	Valleyvet.com
Doramectin, $/mL	$0.32	Valleyvet.com
Fenbendazole, $/mL	$0.10	Valleyvet.com
Tildipirosin (metaphylaxis), $/mL	$3.96	Valleyvet.com
BRD pull charge,3 $/n	$3.00	Assumed
First BRD treatment,4 $/mL	$0.87	Valleyvet.com
Second BRD treatment,5 $/mL	$0.56	Valleyvet.com
Third BRD treatment,6 $/mL	$0.15	Valleyvet.com
Render fee,7 $/n	$32.85	Sparks Companies Inc. (2002)
Feed and yardage, $/dry ton	$249.40	CattleFax
Interest,8 fixed yearly %	7.00	Federal Reserve Bank of KC—Ag Credit Survey
Revenue
Dressed fed cattle base price, $/cwt	$196.40	LM_CT154 (USDA-AMS)
Premiums and discounts9		LM_CT155 (USDA-AMS)
Prime, $/cwt	$12.30
Choice, $/cwt	$0.00
Select, $/cwt	−$10.60
Standard, $/cwt	−$30.20
YG 1, $/cwt	$3.90
YG 2, $/cwt	$1.80
YG 3, $/cwt	$0.00
YG 4, $/cwt	−$11.40
YG 5, $/cwt	−$17.80
Heavyweight, $/cwt	−$19.60
Removal base price,10 $/cwt	$111.40	LM_CT168 (USDA-AMS)

¹Hundredweight = cwt = 45.4 kg (100 lb); 1 U.S. ton = 907 kg (2,000 lb).

²Chute charge accounts for equipment and labor; Bovilis Vista 5 SQ (Merck Animal Health, Lenexa, KS); Bovilis Vision 7 with Spur (Merck Animal Health); Revalor-XS (Merck Animal Health); doramectin (Dectomax; Zoetis Animal Health, Parsippany, NJ); fenbendazole (Safe-Guard; Merck Animal Health); tildipirosin (Zuprevo; Merck Animal Health) administered to metaphylaxis, TECH-high, and TECH-low pens.

³BRD pull charge accounts for equipment and labor for both pulling, diagnosing, and treating the animal.

⁴Florfenicol and flunixin meglumine (Resflor gold; Merck Animal Health) administered SQ at 6 mL/45.4 kg.

⁵Enrofloxacin (Baytril 100; Elanco Animal Health) administered SQ at 5 mL/45.4 kg.

⁶Oxytetracycline (Bio-Mycin 200; Boehringer Ingelheim Animal Health, Duluth, GA) administered SQ at 4 mL/45.4 kg.

⁷Estimated cost of rendering a deceased animal ($24.11) reported by Sparks Companies Inc. (2002), inflation adjusted using the Producer Price Index to a value reflective of time of trial conduct.

⁸Yearly rate was adjusted to be reflective of the number of days-on-feed for each pen out of 365, and applied to the purchase cost of cattle, and one-half of the feed and yardage cost.

⁹Adjustment of the base price is made via premiums and discounts, where the prices are multiplied by the corresponding percent of the pen within each category; YG = Yield Grade; heavyweight discount is for carcasses weighing over 746 kg (1,050 lb).

¹⁰Listed price is reflective of the price for dressed Breaker cows weighing over 227 kg (500 lb), and is further adjusted depending on the estimated carcass weight of animals removed from the trial (Horton et al., 2021).

As the pen was the experimental unit in the RCT, a budget was applied at the pen-level to each pen in the trial. The final outcome is a net return, which is defined here as: total pen revenue minus partial pen costs, converted to a per-animal or per-carcass basis for interpretation. Economic variables from the partial budgets then can be analyzed statistically, similar to continuous performance outcomes (e.g., average daily gain), allowing for comparison of marginal differences between treatments. First, we describe the partial budget framework, followed by an example showing all calculations applied to one of the trial pens.

Partial budget framework

All relevant observed cattle health, performance, and carcass data were included in the partial budgets, regardless of their statistical significance reported by Nickell et al. (2021). This approach is recommended because most RCTs are designed to detect treatment effects on a primary outcome of interest (e.g., BRD morbidity, carcass weight) and may not have sufficient statistical power to detect significance for secondary outcomes, risking Type II error (Esposito et al., 2009). Researchers must exercise caution when interpreting nonsignificant P-values, as the absence of evidence is not evidence of absence (Bello and Renter, 2018). Excluding variables from an economic analysis due to a lack of statistical significance may introduce bias if Type II error is present, and the researcher falsely concludes that treatment effects were equivalent when they are in fact not. Therefore, we do not recommend using P-values as inclusion criteria for variable selection in a partial budget or other economic assessments. Even if certain production variables do not show statistical significance, they may still have substantial economic significance, particularly when aggregated to a total dollar value for a pen. Although only items that have the potential to be impacted by the change need to be included in the partial budget (Corbin and Griffin, 2006), P-values should not be solely relied upon to determine potential or suspected impacts.

Dixon et al. (2022) noted insufficient reporting of parameter values and sources in their review of economic assessments in feedlot cattle trials. All or some of the values used in economic assessments were reported in 86% of trials, but only 57% reported the sources of these values, and only 39% reported the dates when values were estimated or sourced. The prices (USD) used in the partial budget were reflective of prices at the time of trial conduct (September 2018 to March 2019). This was done out of simplicity for our example. Alternatively, short- or even long-run price averages could be used and may be more appropriate. Seasonal considerations also could be made depending on the producer or production scenario. For example, if the target population is weaned feeder calves typically sold in the fall, using a fall price average may be appropriate, whereas for fed cattle sold year-round, yearly averages may be more appropriate. These are considerations that must be made, then clearly reported and justified by the researcher. Additionally, contemporary prices could be applied to data from historical RCTs, which may be appropriate if researchers are comfortable with the assumption that animal populations, management strategies, and effect estimates remain valid. A complete list of prices used for calculating pen-level costs and revenues in this example, and their sources, are reported in Table 1.

Feeder steer purchase prices were sourced from the Livestock Marketing Information Center (LMIC), which compiles reports from feeder cattle auctions reported by the U.S. Department of Agriculture (USDA) Agricultural Marketing Service (AMS). Weekly prices for medium and large frame #1 feeder steers weighing 272 to 295 kg (600 to 650 lb)—reflective of the average initial body weight (BW) of calves enrolled in the trial—were obtained for each enrollment block, based on enrollment date and origin (Missouri, Oklahoma, Texas) of the trial cattle. An average feeder steer price of $163.70/cwt (hundredweight; cwt = 45.4 kg = 100 lb) was calculated. A processing regimen that included routine vaccinations, anthelmintic and parasitic control, identification, and implanting was constructed to reflect the products administered in the RCT. Apart from an assumed $1.50/animal chute charge, which accounts for equipment and labor, all other prices for pharmaceutical products were sourced from Valleyvet.com after the trial had been completed and are listed in Table 1. Note that antimicrobial metaphylaxis was administered only to META pens and selected animals within TECH-high and -low pens. Some of these processing products are administered on a per-animal (or per-dose) basis, while others are priced per milliliter, with the amount administered calculated based on pen weight and manufacturer dosage instructions. Differing antimicrobials were administered depending on how many times an animal had been treated for clinical BRD, and their prices, again sourced from Valleyvet.com, are in Table 1. Additionally, each time an animal within a pen was treated for clinical BRD, a $3.00/animal pull-and-treat charge was applied, accounting for equipment use and labor to pull, diagnose, and treat individual animals. The main impact of feedlot mortalities is the absence of revenue from the purchased animal and subsequent inputs (e.g., consumed feed). Additionally, a $32.85 rendering fee was applied, adjusted for inflation to 2019 USD from a 2002 reported value ($24.11; Sparks Companies Inc., 2002) using the Producer Price Index (U.S. Bureau of Labor Statistics).

The feed and yardage price ($249.40 per ton of dry matter; 1 U.S. ton = 907 kg = 2,000 lb) was calculated from the CattleFax ration prices database for the Central Plains region of the U.S. This value represents an average reflective of the months during which the steers were on feed in the trial. As all cattle in the trial were fed common diets, this was deemed a reasonable approach. In trials comparing different dietary formulations, ingredients, or management practices, it may be necessary to determine prices for specific diets and apply those prices to specific amounts fed of each. In feedlots, yardage accounts for overhead costs including labor, equipment, and utilities. CattleFax reports industry average feed and yardage prices from commercial beef feedlots monthly. As different pricing structures for feed, feed markup, and yardage are used (e.g., some feed markup and some yardage, yardage and no feed markup, or feed markup and no yardage), a single combined value for feed and yardage is reported. If researchers desire more precision, prices of individual ingredients can be used to calculate total ration prices. Note, however, that unless prices for all ingredients are readily available, it may be challenging to source prices for certain ingredients (e.g., micro-ingredients or others used in small quantities).

The final cost variable included in the partial budget was an interest charge, for which an average reported rate of 7% for Oklahoma during the trial period was used. This rate was sourced from the Federal Reserve Bank of KC—Ag Credit Survey of yearly fixed interest rates for operating loans. When constructing budgets, interest often is applied to certain investments to account for capital use and opportunity costs (Rushton, 2007). Interest commonly applies to fixed and variable expenditures that tie up money over time. For instance, in feedlot cattle production, interest might be applied to the full purchase price of feeder cattle, because this investment occurs at the start of the feeding period and remains outstanding until marketing. Conversely, feed is purchased and consumed gradually, so interest is typically applied to half of the total feed costs, based on the assumption that, on average, half of the feed investment is outstanding at any given time. Closely related is the concept of discounting, where future revenues and costs are adjusted (using interest rates) back to their present value to reflect the time-value of money—recognizing that a dollar received today is worth more than one received in the future due to potential returns from investments (Rushton, 2007). Although discounting may be less critical over short production cycles, it can significantly influence economic outcomes when interventions extend the timing of marketing or delay when revenues are realized (Horton et al., 2024, 2025; Schmaltz et al., 2024). Therefore, interest was applied to both the purchase cost of feeder cattle for the time they were on feed and half of the feed cost, using simple, noncompounding interest.

Revenue in the partial budget was derived from fed cattle sales and removals (i.e., animals culled or “railed” from the trial for health reasons and sold individually at a discounted price). Here, we considered only marketing fed cattle on a dressed (carcass) basis with a premium and discount-based grid for carcass characteristics (Table 1). In some cases, there may be a benefit to considering multiple sale bases (e.g., live marketing). The dressed-fed cattle base price ($196.40/cwt) was sourced from the National Weekly Direct Slaughter Cattle—Negotiated Purchases report (LM_CT154; USDA-AMS). This price reflects steers grading over 80% Choice, sold dressed and delivered on a negotiated cash basis, and was calculated based on the average price of steers sold during the timeframe when trial cattle were harvested in the trial. The base price was then adjusted for premiums and discounts for USDA Quality Grade and Yield Grade (QG and YG, respectively), as well as a discount for heavyweight carcasses [those weighing over 477 kg (1,050 lb)]. There were no cattle in the trial weighing less than 249 kg (550 lb); therefore, no lightweight discount was included. Values for these premiums and discounts are reported in Table 1 and were sourced from the National Weekly Direct Slaughter Cattle—Premiums and Discounts report (LM_CT155; USDA-AMS), again averaging the timeframe when trial cattle were harvested.

The USDA-AMS sets Choice and YG 3 to $0.00/cwt, meaning no premium is given to carcasses grading Choice in the partial budget. While this likely impacts total revenue, the Choice-Select spread remains intact, and it does not impact marginal revenue or net return differences between treatment groups, which is the objective of the partial budget. Additionally, as very few cattle in the trial graded sub-Select, those that did were combined into a single “other” category (Nickell et al., 2021), and the QG discount for carcasses grading Standard was applied to these carcasses. For simplicity in our example, the final grid price was calculated per Feuz et al. (2003), where the premiums and discounts were multiplied by the proportion of carcasses in each category for each pen, and the net adjustment was added to (or subtracted from) the dressed fed cattle base price. Alternatively, one could multiply the grid prices by the respective cumulative weight of carcasses in each category (if those data are available) to determine dollar adjustments for premiums and discounts for pen revenue from fed cattle sales. Anecdotally, when comparing these 2 methods for grid pricing with RCT data, we have found very similar to equivalent results, with approximately a 0.25% difference between pen revenue calculations on average.

The last component of the partial budget involved estimating revenue for animals removed from the trial, for which there is no publicly available source for obtaining these prices. Horton et al. (2021) reported that prices received for feedlot culls were most correlated with cull cow prices—specifically dressed Breaker cows weighing over 227 kg (500 lb)—and recommended a corresponding adjustment of the cull cow price. As removals could occur at any point in the trial, an average cull cow price ($111.40/cwt dressed) was calculated from the trial timeframe using the National Weekly Direct Cow and Bull Report—Negotiated Price report (LM_CT168; USDA-AMS). To adjust the price to estimate removal revenue, the hot carcass weight (HCW) of removals needed to be estimated, for which the inverse of a formula from Tatum et al. (2012) was used, where HCW = 0.2598 × live BW^1.1378. The mean live BW of steers at the time of their removal for a pen was used to estimate the mean HCW of removals from that pen. Then, the cull cow price was adjusted by multiplying it by 0.595, 0.763, or 0.92 when the mean estimated HCW of removed animals for a pen was less than 181 kg (400 lb), between 181 and 271 kg (400 to 599 lb), or greater than 272 kg (600 lb), respectively (Horton et al., 2021).

Calculations and partial budgeting analysis

With all prices used in the partial budget and their sources identified (Table 1), we demonstrate how the budget can be applied to an example pen (experimental unit) from the trial, showing all calculations (Table 2). The pen chosen was from the positive control (META) treatment, with basic demographic and performance metrics including 71 steers enrolled, mean initial BW 262 kg, 260 d-on-feed, 12 first-time BRD treatments, 4 second-time BRD treatments, 2 third-time BRD treatments, 0 removals, 1 mortality, 7.85 kg/d mean dry matter intake, 70 steers shipped and harvested, and 402 kg mean HCW. Note that pen-level totals (e.g., total kg HCW, dry feed delivered) are used in the budget (Table 2), not mean values, and that background conversion of units (e.g., cwt to kg) is not shown. Notably, Table 2 appears like a gross margin analysis for a single pen, excluding constant or fixed costs unaffected by treatment. Its purpose is to show how pen-level calculations were performed. The partial budget is formed when the data are analyzed using statistical models (described later), where the models estimate the marginal treatment effects of the interventions on net returns (or other economic outcomes of interest), using the results of the budget when applied to each pen from the RCT.

Table 2.

Example budget applied to one trial pen¹

Costs	Price	Unit2		Quantity	Unit		Value
Cattle purchase	$163.70	per cwt	×	18,575	kg	=	$67,036.62
Per n processing3	$12.16	per n	×	71	n enrolled	=	$863.36
Doramectin4	$0.32	per mL	×	372	mL	=	$119.04
Fenbendazole5	$0.10	per mL	×	942	mL	=	$94.20
Tildipirosin6 (metaphylaxis)	$3.96	per mL	×	410	mL	=	$1,623.60
BRD pull charge	$3.00	per n	×	18	n treated	=	$54.00
First BRD treatment7	$0.87	per mL	×	428	mL	=	$372.36
Second BRD treatment8	$0.56	per mL	×	114	mL	=	$63.84
Third BRD treatment9	$0.15	per mL	×	54	mL	=	$8.10
Render fee	$32.85	per n	×	1	n mortality	=	$32.85
Dry feed and yardage	$249.40	per ton	×	144,385	kg	=	$39,693.84
Interest on feed10	2.49	%	×	39,693.88	$	=	$989.63
Interest on cattle11	4.99	%	×	67,035.15	$	=	$3,342.57
Total costs12							$114,294.02
Premiums and discounts13
Prime	$12.30	per cwt	×	8.6	%	=	$1.06
Choice	$0.00	per cwt	×	80.0	%	=	$0.00
Select	−$10.60	per cwt	×	11.4	%	=	−$1.21
Standard	−$30.20	per cwt	×	0.0	%	=	$0.00
YG 1	$3.90	per cwt	×	11.4	%	=	$0.44
YG 2	$1.80	per cwt	×	38.6	%	=	$0.69
YG 3	$0.00	per cwt	×	41.4	%	=	$0.00
YG 4	−$11.40	per cwt	×	8.6	%	=	−$0.98
YG 5	−$17.80	per cwt	×	0.0	%	=	$0.00
Heavyweight	−$19.60	per cwt	×	1.4	%	=	−$0.27
Total price adjustment							−$0.27
Grid price	$196.40	per cwt	+	−$0.27	per cwt	=	$196.13
Revenue
Fed cattle sales	$196.13	per cwt	×	28,164	kg	=	$121,781.64
Removal sales	-	per cwt	×	0	kg	=	$0.00
Total revenue							$121,781.64
Net return
Pen total	$121,781.64	Revenue	–	$114,294.02	cost	=	$7,487.63
Per animal enrolled	$7,487.63	Return	÷	71	n enrolled	=	$105.46
Per carcass	$7,487.63	Return	÷	70	n carcasses	=	$106.97

¹Pen was in the negative control (metaphylaxis) treatment, and had 71 steers enrolled, mean initial BW 262 kg, 260 DOF, 12 first time BRD treatments, 4 second BRD treatments, 2 third BRD treatments, 0 removals, 1 mortality, 7.85 kg/d mean dry-matter-intake, 70 steers shipped and harvested, 590 kg mean shrunk final BW, 402 kg mean HCW.

²Hundredweight = cwt = 45.4 kg (100 lb); 1 U.S. ton = 907 kg (2,000 lb).

³Sum of processing product prices that were delivered on a per animal (dose) basis: $1.50 chute charge; $0.39 ear tag; $1.17 Bovilis Vista 5 SQ (Merck Animal Health, Lenexa, KS); $0.65 Bovilis Vision 7 with Spur (Merck Animal Health); $8.44 Revalor-XS (Merck Animal Health).

⁴Administered SQ at 1 mL/50 kg (Dectomax; Zoetis Animal Health, Parsippany, NJ).

⁵Administered orally at 2.3 mL/45.4 kg (Safe-Guard; Merck Animal Health).

⁶Administered SQ at 1 mL/45.4 kg (Zuprevo; Merck Animal Health).

⁷Florfenicol and flunixin meglumine (Resflor gold; Merck Animal Health) administered SQ at 6 mL/45.4 kg.

⁸Enrofloxacin (Baytril 100; Elanco Animal Health) administered SQ at 5 mL/45.4 kg.

⁹Oxytetracycline (Bio-Mycin 200; Boehringer Ingelheim Animal Health, Duluth, GA) administered SQ at 4 mL/45.4 kg.

¹⁰Interest rate for interest on feed is half of the fixed interest rate multiplied by the number of days-on-feed divided by 365: (7%/2) × (260/365).

¹¹Interest rate for interest on feeder cattle purchase is the fixed interest rate multiplied by the number of days-on-feed divided by 365: 7% × (260/365).

¹²Total costs in the budget are total partial costs for the enterprise (i.e., constant or fixed costs unaffected by treatment are excluded from the budget).

¹³YG = Yield Grade; heavyweight discount is for carcasses weighing over 746 kg (1,050 lb).

The cost of cattle purchased was calculated by multiplying the feeder steer price by the total initial weight of the pen (Table 2). Processing products administered on a per-animal or per-dose basis (as listed in Table 1) were summed to a total price per animal and termed “per n processing” in Table 2. This price was multiplied by the number of animals enrolled in the pen. Processing products and clinical BRD treatment products priced on a $/mL basis were multiplied by the number of milliliters delivered to animals, determined by manufacturer-labeled dosages and the mean weight of the pen at processing or the actual weight of individual animals when treated for BRD. The BRD pull charge and render fee were multiplied by the total number of clinical BRD treatments and mortalities, respectively, in the pen. The feed and yardage price was multiplied by the total weight of dry feed delivered to the pen. Interest was applied to both feed and cattle by converting the yearly interest rate (7%) to the actual interest based on the days-on-feed for the pen and multiplying it by the total cattle and feed costs. To account for interest on one-half of the feed cost, half of the interest rate was used in this calculation.

Before proceeding to revenue calculations, the grid price was determined by adjusting the dressed-fed cattle base price according to premiums and discounts and the frequencies of quality grades and heavyweight carcasses in the pen (Table 2). The premium or discount for each category was multiplied by the percentage of animals in the pen within each respective category to determine the adjustments. The adjustments were then added to the base price to calculate the grid price. Revenue from fed cattle sales was calculated by multiplying the grid price by the total HCW of the pen (Table 2). As previously described, the price for cattle removed from the trial would vary between pens, depending on the weight of the removed animals. As there were no removed animals in the example pen, no price or revenue is listed in Table 2 for removals.

The final net return calculation is the total pen revenue minus the sum of partial pen costs (Table 2). This value can then be converted to a per-animal enrolled or per-carcass basis. We note that while this may seem similar to “deads-in” vs. “deads-out” calculations of animal performance variables, all economic calculations were performed with all dead and removed animals included. Therefore, net return per-carcass calculations would still be interpreted as deads-in. We present both denominators here to show that, in most cases, the difference is minimal and may be up to the preference of the researcher or producer. Table 2 is meant to be an example of how all prices and calculations can be applied at the pen-level in an RCT of feedlot steers. We note that a table like Table 2 is not created for each pen in the trial for calculations. Rather, this process is streamlined by conducting the calculations across rows (experimental units) in the dataset, using the researchers’ software of choice.

Finally, to execute the partial budget, statistical analyses were performed on the results of the budget applied to each pen, similar to the analysis of continuous performance variables in the trial (Nickell et al., 2021). Linear mixed models (Proc GLIMMIX, SAS 9.4; SAS Institute Inc., Cary, NC) were fitted with a Gaussian distribution, an identity link function, a Kenward-Roger degrees of freedom adjustment, Newton–Raphson and Ridging optimization procedures, and used restricted maximum likelihood estimation. Pen was the experimental unit, treatment was the fixed effect, and block was included as a random intercept. Model assumptions of normally distributed and homogeneous residual variances were assessed visually using plots of studentized residuals. A Tukey–Kramer adjustment for multiplicity was used for pairwise comparisons between treatments. The primary outcome was net return; however, other variables within the partial budget (e.g., cost of gain) were also included for evaluation. An a priori significance threshold was set at α = 0.05.

Results and reporting

Results of the partial budget analysis are presented in Table 3, with NEG set as the referent treatment and the marginal differences of the other treatments compared to NEG reported. As anticipated, there was no evidence of a difference in the purchase cost of steers (P = 0.52), due to the fixed feeder cattle price, and the randomization process resulting in similar initial BW between treatments (Nickell et al., 2021). Arguably, steer purchase costs could have been excluded from the partial budget since randomization of steers to treatments should result in BW, and thus purchase costs, that are not associated with experimental treatments. However, there is no harm in including additional variables in partial budgets, and mean initial BW varied numerically across individual pens and treatments, implying the potential for differing pen-level cattle purchase costs. Additionally—as previously mentioned—it is not recommended to rely on statistical significance for variable inclusion. Processing costs differed among all treatments (P < 0.01), reflecting the amount of metaphylactic doses administered to the pens. The pen-level cost of treating animals with clinical BRD was affected by treatment (P = 0.02), with META and TECH-high having lower costs per animal enrolled compared to NEG. Although TECH-high had higher feed and yardage costs compared to NEG (P < 0.01), there was no evidence of a treatment effect on the cost of gain (P = 0.14). Metaphylaxis and TECH-high had higher total pen costs per animal than NEG (P < 0.01), while TECH-low did not differ significantly from any of the treatments. Total revenue was affected by treatment (P = 0.02); however, only TECH-high differed significantly from NEG by receiving more total revenue per animal, with no evidence of differences among the other treatments. Evaluating net return on a per-animal enrolled or per-carcass basis yielded similar results, both finding no significant differences between treatments (P = 0.34 and 0.31, respectively).

Table 3.

Model-adjusted differences between the means and standard errors of the differences (SED) from a partial budget assessment of a randomized controlled trial comparing the effects of standard BRD control programs with targeted programs for beef feedlot steers

	Treatment1
Item	NEG	META	TECH-high	TECH-low	SED	P-value
Purchase cost,2 $/n enrolled	Referent	4.55	−0.87	1.04	3.798	0.52
Processing cost,3 $/n enrolled	Referenta	24.64b	21.29c	15.26d	1.073	<0.01
Clinical BRD treatment cost,4 $/n enrolled	Referenta	−4.24b	−4.07b	−3.40ab	1.296	0.02
Feed and yardage cost,2 $/n enrolled	Referenta	20.93ab	33.65b	10.12ab	8.748	<0.01
Cost of gain,5 $/cwt	Referent	−5.24	−3.63	−3.39	2.145	0.14
Total cost, $/n enrolled	Referenta	45.62b	49.81b	23.02ba	9.627	<0.01
Total revenue, $/n enrolled	Referenta	76.84ab	86.61b	20.50ab	28.924	0.02
Net return, $/n enrolled	Referent	31.21	36.81	−2.51	26.664	0.34
Net return, $/carcass	Referent	31.67	38.29	−4.55	27.261	0.31

¹Negative control (NEG; no metaphylaxis), a positive control [META; tildipirosin (TIL) metaphylaxis], precision technology (TECH) TECH-high (TIL administered to individual animals based on a conservative algorithm threshold), and TECH-low (TIL administered to individual animals based on an aggressive algorithm threshold).

²Includes interest.

³Cost of all products given at initial processing, including TIL, with the amount dependent on the treatment.

⁴Cost of all clinical BRD treatments administered to animals within a pen, divided by the number of animals enrolled in the pen.

⁵Feed, yardage, and interest cost per cwt live weight gain (cwt = 45.4 kg = 100 lb).

^a,b,c,dUncommon superscript letters within row indicate significance (P ≤ 0.05) after adjustment for multiple comparisons.

The reporting in Table 3 aligns with the objective of a partial budget, which is to evaluate the marginal differences between interventions or management strategies. Therefore, differences between the means (compared to NEG) are shown, rather than the means themselves. If preferred, it would still be acceptable to report model-adjusted means, as the treatment differences would remain identical. However, this comes with some caveats. Firstly, we are not necessarily interested in whether the trial cattle were profitable overall, as this depends largely on the input prices used in the partial budget. Rather, we are interested in how much additional return was gained or foregone by using an alternative intervention, regardless of whether net returns were positive or negative on average. Additionally, since only variables that potentially differ among interventions need to be included in a partial budget, reporting mean values for profitability may be misleading unless the budget is comprehensive (i.e., includes all possible economic variables, even those shared among treatments). Although our example budget presented here was relatively comprehensive and, if reported, the model-adjusted means may reasonably represent overall profitability under the given conditions, it would still lack the detail provided by whole farm or enterprise budgets. Finally, we note that the cost of using the TECH was not accounted for in the partial budget, meaning that for it to be economically advantageous, its cost would need to be less than the marginal improvement in net return.

Discussion and future guidance

The partial budget analysis using RCT data effectively demonstrated the 4 components of partial budgeting, particularly when comparing NEG with TECH-high. Steers in the NEG treatment experienced reduced costs but also reduced revenue, while TECH-high steers had increased revenue accompanied by increased costs. The combination of these components resulted in nonsignificant net return differences between the treatments. It should be acknowledged that although there were no significant differences in net return, mean dollar values exceeding $30 per animal, as observed here, would certainly carry substantial economic significance for producers. This highlights an inherent challenge when performing economic analyses in this manner: net return is a derived or composite variable that relies on a multitude of cattle production and economic variables—each with their own variability—for calculation. This suggests that, in some cases, economic analyses conducted on RCT data would benefit from larger sample sizes with greater statistical power. Nevertheless, researchers should not be discouraged by this observation and should not forgo economic analyses; such analyses are a critical part of the decision-making process for industry stakeholders and should not be left obscure (Antle and Wagenet, 1995; Cernicchiaro et al., 2022; Dewsbury et al., 2022).

Cattle and economic variables included in a partial budget can (and should) vary on a case-by-case basis, depending on specific research objectives and experimental designs of RCTs. In other words, there is no one-size-fits-all budget, and researchers should adapt the partial budget framework to fit specific needs. For example, if the primary objective of an RCT is health-focused (e.g., morbidity), where differences among treatments are expected, it is advisable to distinguish health costs for differing diagnoses and clinical treatment regimens (e.g., Valencia et al., 2019; Nickell et al., 2021; Horton et al., 2023). Conversely, if a research objective is production-focused without hypothesized health effects, it may be reasonable to use a fixed cost for health events rather than extensive differentiation (Horton et al., 2024). Alternative research questions and study designs may also require more complex considerations, such as discounting or the opportunity cost of money (Horton et al., 2024). Therefore, our example partial budget analysis is meant to provide guidance as a framework, which should be tailored to fit specific research questions, hypotheses, and study designs. Other variables and sources for prices can certainly be used, as long as they are economically sound and well justified. In complex situations, it may be wise to consult with an economist when there is uncertainty regarding variable inclusion, market conditions, or other economic factors that may not be immediately apparent. Ideally, this should occur before trial initiation, to ensure that all necessary data for economic analyses are collected.

The primary limitations of partial budget analyses are that they are generally static and do not account for the variability that occurs from market fluctuations affecting prices or, typically, the variability of biological responses such as animal health and performance due to genetics, management, and environmental factors. In their review, Dixon et al. (2022) found that just 44% of trials used a statistical evaluation for economic or financial outcomes. However, in addition to providing more evidence for effects, by performing statistical analysis on a partial budget of RCT data, the second aforementioned limitation is somewhat mitigated because the variability of cattle responses in the trial data is accounted for, and this variability is reflected in the standard errors from the statistical models. The same limitations that apply to any RCT still hold—that is, the scope of inference is relative to the animal population, environment, management, and overall external validity of the trial. However, this approach is still highly advantageous over a traditional partial budget that would use fixed parameter values for animal health and performance variables (i.e., using treatment means for animal variables), due to the inclusion of biological variability. Nevertheless, partial budgets are most commonly a short-term, “snapshot” analysis under a specific set of fixed economic conditions that assume constant external market factors, and they do not fully account for the economic risk that may occur under long-term, fluctuating conditions.

In addition, even well-conducted partial budgets derived from RCT data have inherent limitations regarding their representativeness of future or broader market conditions. Because prices for inputs and outputs can vary significantly across time, regions, and production systems, the economic results observed may not directly translate to other settings unless those external variables remain relatively similar. For instance, a dramatic shift in feed costs or carcass premiums and discounts could substantially alter net returns, even if the biological performance differences between treatments stay the same. Consequently, generalizing findings to different operations or economic climates often requires revisiting the partial budget with updated price assumptions or additional scenario analyses. This highlights why we advocate for transparent reporting of price sources, timeframes, and detailed calculations, so readers can tailor or reevaluate the economic analyses to reflect their particular market environment.

Generalizability and representativeness are also affected by the primary goal of the analysis (e.g., a robust economic risk assessment vs. an analysis to supplement animal health and performance outcomes) and by the biological variability or expected response to the intervention (e.g., a hormonal implant vs. a direct-fed microbial). Again, external validity is limited to the animal population, environment, management conditions in which the RCT was conducted (similar to how primary outcomes are interpreted), and the specific economic conditions used in the analysis. In different economic conditions, there is not a straightforward “treatment vs. control” scenario like in RCTs, so generalizability of economic outcomes to future animals becomes more intricate. Therefore, readers should interpret results accordingly, and authors can facilitate appropriate extrapolations through transparency and clarity in writing, and highlighting any key considerations likely to impact the findings.

There are potential methods to mitigate concerns about the short-term, static nature of traditional partial budgets and their external validity. For one, researchers do not need to feel confined to the economic conditions occurring at the time of an RCT. While we used contemporaneous prices in our example, we could have alternatively evaluated the economic effects of the BRD management interventions using prices from, for example, spring 2024, average prices from 2021 to 2024, or even longer-run inflation-adjusted prices, and compared the results of the budgets constructed under differing economic conditions. This would allow for the evaluation of economic implications under different scenarios, which may better inform industry stakeholders. However, this approach comes with the caveat that one must feel comfortable with the assumption that animal populations and treatment responses from an RCT can be extrapolated and still reasonably represent populations that would align with alternative economic conditions.

When the primary research question is economic, accounting for risk and market variability becomes more critical, and static partial budgets are less likely to suffice. Still, economic risk-assessment methodologies can adopt the same partial budgeting framework. Horton et al. (2024) provide an example of a sensitivity analysis conducted within a partial budget framework to account for risk by evaluating how changes in key economic factors affect economic outcomes, thereby providing a range of possible results. In addition to sensitivity analyses, stochastic modeling may provide even more robust assessments of how interventions perform under varying conditions in order to characterize risk (Corbin and Griffin, 2006), which also may be built upon a partial budget framework (Horton et al., 2025). These techniques will always outperform a static partial budget by producing broader ranges of possible outcomes. However, their necessity is dependent on the primary objective at hand, and the specific intervention(s).

Certain treatment or management interventions may have simpler, more consistent effects that are not heavily influenced by economic conditions, while that may not be the case for others. For example, Horton et al. (2024) found relatively consistent net return differences in terms of direction and magnitude for alternative implant programs in feedlot heifers, regardless of fluctuating economic factors. This is a case where a static partial budget could suffice, as there were consistent and expected biological differences between the implant programs that provided a unidirectional economic benefit regardless of the market conditions. Conversely, the authors found that net returns were heavily influenced by differing economic conditions when extending feedlot heifer DOF, resulting in contrasting interpretations that were dependent on input prices. Interventions that yield variable biological responses, change marketing considerations (e.g., changing marketing windows), or have long-run production effects (e.g., a genetic change in a cow-calf herd), will benefit from—if not require—a more robust economic risk assessment. In summary, strategies exist to mitigate the primary limitations of partial budgets, and the research design and specific intervention in question should determine whether more complex methods are necessary.

An additional limitation of partial budgets, which is of lesser consequence under the context of RCTs but should still be considered, is the exclusion of fixed costs and opportunity costs (Rushton, 2007). Partial budgeting often focuses solely on variable costs and benefits directly affected by the intervention, neglecting fixed costs shared among treatments and the opportunity costs of resources utilized (e.g., time, labor, capital, land). Under the RCT setting, where fixed costs and resource allocations are consistent across treatments, this limitation is minimized. However, when applying results to broader production systems or evaluating long-run scenarios, variations in fixed costs and alternative uses of resources could influence the overall economic impact of an intervention. Therefore, while this limitation has a reduced effect within the controlled trial framework, researchers should still consider these factors when interpreting partial budget analyses and their potential implications in real-world scenarios.

There is a need for more consistent conduct and reporting of economic analyses in livestock research (Saatkamp et al., 2016; Dixon et al., 2022). The lack of clear economic evaluations hampers stakeholders’ ability to make informed decisions based on research findings, and poor reporting practices limit the potential for meta-analyses or systematic reviews, which rely on consistent and comprehensive data. Reporting guidelines for animal research exist, many of which are compiled on the MEnagerie of Reporting guIDelines Involving Animals (MERIDIAN; https://meridian-network.org/) website. Additionally, reporting guidelines for economics conducted in human health research are available, such as the Consolidated Health Economic Evaluation Reporting Standards [CHEERS; Husereau et al. (2013)]. However, none of these guidelines directly apply to the conduct of economic analyses in animal research, particularly for the production of livestock species. The objective of this publication was not to create a detailed and comprehensive reporting guidelines checklist like those found in the previously mentioned sources. However, there are 4 basic principles that we strongly encourage be followed, in order to promote more transparent and reproducible economic analyses of RCT data, so that methodologies and findings may be better interpreted and critically evaluated by the reader.

1. Explicitly state and justify the type of economic assessment: the type of economic assessment performed should be explicitly stated and justified to fit the specific research question at hand.

2. Provide sources and descriptions of all parameters (prices and variables): sources and descriptions of all prices and other variables should be provided, including the timeframe they reflect.

3. Detail the methodology: a detailed description of the methodology should be provided, including how prices were applied and at what level(s) (e.g., animal-level, pen-level).

4. Describe the analytical methods: a description of how analyses for interpretation were performed, including whether statistical tests were used, along with justification of said analyses.

In our case study, the partial budget framework allowed explicit descriptions of principles 1 through 4, yielding high transparency and reproducibility of the findings. Along with these principles, there is an additional consideration that may impact the quality of economic assessments on RCTs. Like other primary study outcomes, the analysis and methodology should likely be planned a priori, being clearly conceived and described in study protocols. In other words, the type of assessment, variable descriptions and timeframes, methodological processes, and analytics can all be determined before starting a study. This further promotes transparency and dissuades “data fishing.” If skipping this step, researchers also may find that the necessary data for thorough economic analyses were not collected, leading to lower quality assessments. This again relates to the benefit of collaboration with economists, particularly as the complexity of economic implications or scenarios rises.

We have recommended and demonstrated the use of partial budgets for the economic assessment of RCT data, as the purpose of partial budgets is generally well aligned with the evaluation of alternative interventions or management strategies in RCTs. However, this does not mean that other methods are unacceptable; there may be cases for RCTs where, for example, a cost-benefit analysis may be more appropriate when the goal of the researcher is to incorporate an evaluation of nonmonetary costs and benefits. Additionally, we have recommended the use of traditional statistical analyses where budgets are constructed at the experimental unit level in an RCT, as this incorporates biological variability into the estimates. Still, there may be cases where statistical analyses become less appropriate, such as with stochastic simulation modeling where sample sizes may be artificially increased, and the objective is rather to characterize economic risk between interventions. Partial budgets provide crucial but context-dependent insights, and researchers should adjust them for fluctuating prices or consider more advanced methods if long-term effects or highly variable market conditions are expected. Ultimately, the economic analysis type, price and variable sources, methodology, and analyses should be well justified and described by the researcher, so that readers and stakeholders may make better-informed decisions based on robust and transparent economic evidence.

Conclusions

Including economic outcomes in research involving production livestock and RCTs in particular, is critical for providing a comprehensive evaluation of interventions and management strategies. While our examples were in the context of feedlot cattle research, there is no reason that the principles outlined here cannot be applied to other cattle sectors or production livestock species. Economic analyses offer valuable insights into the financial viability of alternative interventions or management strategies, which are essential factors for decision-making by producers and other industry stakeholders. By integrating economic outcomes with common outcomes in RCT data, researchers can present a more holistic view of the implications of their findings, thereby enhancing the practical applicability and impact of their research.

Thorough and robust reporting of economic analyses is imperative to ensure transparency, reproducibility, and credibility of the research. Partial budgeting emerges as a suitable and effective method for conducting economic assessments in RCTs due to its alignment with the evaluation of marginal changes associated with alternative treatments or interventions. When accompanied by appropriate statistical analyses, partial budgets can account for biological variability and provide more reliable estimates of economic impact. Still, the generalizability of static budgets should be interpreted accordingly, and researchers should contemplate whether the need to integrate more robust methodologies is warranted based on the RCT and primary research question. Ultimately, economic considerations are just another piece of the puzzle; they should be weighed alongside other important factors such as animal welfare, antimicrobial stewardship, and environmental and social sustainability. By embracing comprehensive and well-reported economic analyses, researchers can contribute to more informed and balanced decision-making processes that ultimately benefit the livestock industry as a whole.

Glossary

Abbreviations:

AMS

Agricultural Marketing Service

BRD

bovine respiratory disease

BW

body weight

cwt

hundredweight

DOF

days-on-feed

HCW

hot carcass weight

LMIC

Livestock Marketing Information Center

META

positive control treatment

NEG

negative control treatment

QG

USDA Quality Grade

RCT

randomized controlled trial

TECH

precision agriculture technology

USDA

United States Department of Agriculture

YG

USDA Yield Grade

Conflict of interest statement. D.R. and D.P. have received previous funding from Merck Animal Health. No competitive interest was present due to non-use or evaluation of competing products. There are no conflicts of interest to disclose.

Author contributions

Lucas Horton (Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Writing—original draft), David Renter (Conceptualization, Data curation, Funding acquisition, Project administration, Resources, Supervision, Writing—review & editing), and Dustin Pendell (Conceptualization, Formal analysis, Methodology, Project administration, Supervision, Writing—review & editing)