— Merck Animal Health warns caution, says study is not scientifically rigorous
When it comes to disputes of a factual nature, the referee comes wearing solid white rather than zebra stripes. But problems arise when two lab-coated refs show up claiming to have seen vastly different plays, and arguments over whose science is best are never pretty.
Researchers out of Texas Tech University and Kansas State University, led by Dr. Guy Loneragan, recently released a report on a collection of studies conducted on the effects of beta agonist use in feedlot cattle. The report, “Increased mortality in groups of cattle administered the -Adrenergic Agonists ractopamine hydrochloride and zilpaterol hydrochloride,” was published on the free-access online academic journal, PLOS One. Their conclusion is that the use of beta agonists in feedlot cattle nearly doubles the risk of death.
On the other hand, Merck Animal Health, producer of Zilmax— the commercial product including the beta agonist zilpaterol—claims Loneragan’s findings are no more than opinions drawn from unscientific observational data. By contrast, Merck points to the dozens of studies conducted to secure Food and Drug Administration (FDA) approval for its product as more reliable, scientific studies.
“Death is a relatively rare event in feedlot cattle,” read the report’s conclusion. “Even so, the data presented herein provide compelling evidence that administration of FDA-approved AA [beta agonists] to cattle increased both the cumulative incidence (risk) and incidence rate of death.”
That evidence, according to the researchers, comes in the form of three independent datasets collected from a number of large cattle feeding operations around the country. The datasets were compiled by the individual feeding operations that had conducted their own controlled field studies and experimental observations then later shared with the researchers. The researchers analyzed the datasets and tried to control for the differences between the different feeding operations’ independent study parameters.
All told, the three datasets analyzed represented 951,511 beef cattle—majority steers—in commercial feedlots. By comparison, Merck touted “more than 30 studies,” representing 65,000 head of cattle, were used to establish its product’s safety.
As the methods behind the study are a point of contention, the unusual nature of the study’s structure deserves an explanation. As mentioned, the study itself was actually analysis of three different datasets composed of data provided by willing feedlots that had conducted their own research on the effects of beta agonists as used in accordance with FDA-approved feeding regimens for the respective products.
The first dataset was composed of data from four different feedlot companies that examined the effect of ractopamine use on their cattle. According to the report, the individual experiments done were randomized and included control groups that were not treated with ractopamine.
“All experiments used a randomized block design in that there were at least two groups per block, i.e., one administered [ractopamine] and the other not. These data included information suitable for analysis on a total of 79,171 cattle.”
The analysis of this dataset showed that overall cumulative risk of death for experiment cattle (both treated and control groups) was 0.26 percent and actual death incidence was .86 deaths per 10,000 “cattle days.”
“Cattle days”—and “animal days” used later in the report—were not directly defined, but from context it can be understood to mean the number of days experienced by the number of cattle in the study. So a pen of 10 cattle watched for three days would collectively represent 30 cattle days.
Analysis of the dataset divided by treatment or control group, however, showed an interesting detail. While the untreated control groups had a 0.18 percent cumulative risk and 0.59 deaths per 10,000 cattle days, the ractopaminetreated experimental groups had a cumulative risk of 0.34 percent and an incidence of death rate of 1.12 deaths per 10,000 cattle days.
“After controlling for clustering within company, study, block and group, cattle administered [ractopamine] were 91 percent more likely to die than control animals during the atrisk [test] period,” summarized the report when discussing the first dataset.
The second dataset was composed of observational data provided by nine different feedlots, with the usable data collectively representing 722,704 cattle. The data collected was on observations of cattle treated with zilpaterol compared to those not treated. The data, however, did not come from balanced studies as in the first dataset.
The researchers noted that the majority (637,339 head) of the cattle represented by the data were in the treatment group, while the remaining minority represented the untreated group. There was no mention of randomization in how cattle were sorted into these groups. This dataset also included information on potential contributing factors to the effects observed. However, only the month of the year cattle were shipped for slaughter (assumed to be proxy for temperature) was found to be statistically significant.
The analysis found again that treated cattle were far more likely to be at risk of death and have higher incidence of death rates than untreated cattle. The researchers concluded that treatment with zilpaterol “was associated with a significant increase in the likelihood of death” and that this risk increased in the warmer months of the year.
The third dataset represented observational data on 149,636 cattle housed at one feedlot for the effect of zilpaterol on death and overall well-being. The cattle were broken into treated and untreated groups more evenly than in the second dataset, but again there was no mention of randomization in this division.
The researchers concluded that treated cattle in the third dataset were 85 percent more likely to die than untreated cattle and far more likely to require treatment—and repeated treatment—for various illnesses.
“Animals administered [zilpaterol] were 23 percent more likely to require treatment for any condition during the exposure period… The likelihood of treatment, however, varied by the body system to which the clinical signs were attributed. Animals were 34 percent less likely to require treatment for conditions attributed to the digestive system…but 2.3 times more likely to require treatment for conditions attributed to the respiratory system. …In addition, animals administered [zilpaterol] were 2.3 times more likely to require more than a single treatment regimen for respiratory disease than contemporaneous controls … .”
[Editor’s note: The more technical numbers, including P-values and confidence intervals, originally present in the above quote were removed for ease of reading. If you would like to read the whole report complete with technical details, the report is available online at plosone.org/article/%3Adoi %2F10.1371%2Fjournal. pone.0091177 ]
From all of the data presented and analyzed, the researchers conclude treatment with beta agonists increases the admittedly low risk of death in feedlot cattle.
“Despite the potential limitations of the data, we argue that given the magnitude of the data, and the strength and consistency of the various measures of effect, both [ractopamine] and [zilpaterol] are most likely causally associated with increased cumulative incidence, incidence rate, and hazard of death when they are administered in accordance with the FDAapproved label directions,” they concluded.
“If so, we believe a broad and inclusive dialogue that explores the balance between improved production efficiencies achieved through means such as AA and resultant adverse effects on the welfare of animals we raise for food is needed. This is particularly warranted for those drugs that are approved solely to improve the efficiencies of production, yet offer no offsetting health benefits to the animals to which it is administered.”
But Merck has different opinions on the issue.
“Only a well-designed clinical study, in a real world setting coupled with analysis of data by third party industry experts can thoroughly confirm the safety profile and performance of a product,” the company asserted in its official response to the report.
“In contrast, Dr. Loneragan’s opinions are based on observational information and we disagree with them. Using observational analyses, where cattle are not randomized and where rigorous scientific procedures are not utilized, is not a respected scientific method to rigorously evaluate the safety and efficacy of any product. For these reasons, caution should be used in drawing conclusions from analyses such as those conducted by Dr. Loneragan.”
Merck pointed out that over 30 studies “conducted by well-respected universities and third party experts using sound scientific principles” found no such increased incidence of death or illness.
“In addition to the 30-plus randomized, controlled studies conducted to date, Merck Animal Health is also conducting an extensive field evaluation program that is being conducted by independent experts as part of our fivestep plan.”
In August of 2013, Tyson announced it would not accept cattle fed zilpaterol over health and welfare concerns for the cattle. Other packing companies quickly followed suit and Merck voluntarily stopped selling the product, announcing the launch of its investigation into welfare allegations.
Interestingly, the Merck response to the report claimed the field evaluation program will support the results of previous studies, though the evaluation is not complete and there was no mention of completion date or why the results might be known prior to finishing the evaluation.
As mentioned earlier, Merck claims the aforementioned studies represented 65,000 head of cattle, and this detail was indirectly addressed in the report.
“Based on the results observed herein, the number needed to harm, a measure of extent of exposure required for a AA-related death, was approximately 500 animals or 15,000 animal days. In most reports of well-controlled cattle experiments, the number of animals included is usually insufficient to speak to an association of AA with mortality, given the rarity of death and estimates of the number needed to harm.”
The researchers asserted the analyses carry considerable strength given the volume of data and the nature of the setting in which the data was collected, i.e., commercial feedlot settings. They claim the real-world setting adds “a degree of validity to the results.” — Kerry Halladay, WLJ Editor