Start with the denominator. Overall Australian feedlot mortality runs under 0.5%, and respiratory disease accounted for nearly 38% of the deaths recorded here [3][4]. Push those through the headcount and you get on the order of 5,200 deaths, about 2,000 of them from BRD, spread across six years and three yards [15]. That is the base the four-fold figure sits on. A consignment from a high-mortality supplier still carries a small risk of dying, one that has been multiplied, and even quadrupled it stays under roughly 1% [16][5].
Which is why the timing curve does more work here than the relative risk. Around 69% of BRD deaths fell inside the first 60 days on feed, peaking near day 35 [7]. Arrival is preceded by mustering, yarding, transport and mixing, and cattle bought through auction yards died of BRD at significantly higher rates than cattle taken straight off a paddock, which the authors read as consistent with that extra handling and commingling [6]. By the time the disease is visible in a pen, the exposure and the stress that produced it are weeks and one truck ride behind.
The design is an analysis of existing health and production records rather than an assigned experiment [17], and that sets what it can carry. Supplier history earns its keep as a screening variable without anyone establishing why a property is risky: genetics, weaning age, haul distance, prior pathogen exposure all remain live candidates, and a buyer sorting consignments does not need to tell them apart. Darren Trott of Adelaide University is explicit that such a profile has to be refreshed as new performance data arrive rather than treated as a fixed property of the farm [10].
The pre-arrival vaccination result deserves more caution. Vaccinated cattle had substantially lower BRD mortality, strongest across the first two months [8], which is exactly where the deaths are, and principal author Gebremeskel Werid argues immunity needs time to develop before transport and induction [11]. In records data, though, producers who vaccinate before loading are also producers who do other things well. The agreement between the vaccine window and the mortality window offers mild support, short of a controlled comparison.
What remains unmeasured is the size of the problem it is being asked to price. The measured outcome is death from BRD [18]. A feedlot's respiratory bill is mostly animals that get pulled, treated, lose weight and recover, and none of those appear in a mortality count. Antimicrobial stewardship is where the authors expect prevention to pay [14], and that case runs on treatments, not carcasses. Breed and season carry the same caveat: Bos indicus and crossbred cattle fared better than temperate breeds [12], and much of the seasonal variation dissolved into heat stress and differences between source properties once those were accounted for [13], which suggests some apparent effects are the source property wearing another hat.
My read: the supplier record is usable now as a buying and sorting rule, because prediction does not require causation. The vaccination timing claim I would hold as a strong prior with an unresolved confound attached.
