PROVE THE VALUE
Analysis by Animal Health Institute of the Human Health Risks of Antibiotics in Animal Feed
Popular concern about alleged health risks posed by the use of antibiotics in animal feed, an agricultural practice that had been in place for more than a half a century, reached a fever pitch in 2001. Recent dramatic assertions by health policy activists and others suggested that such antibiotics would promote the spread of antibiotic-resistant bacteria in food animals and meat products, ultimately to the detriment of human health.
The result, among other things, was a dramatic slowdown in sales of virginiamycin, an antibiotic used to prevent illnesses and promote feed efficiency in cattle, chickens, and pigs. Its use by European Union member states had been banned by the E.U. in 1998. In addition, serious scientific questions were emerging about the potential harm to human health from the suspension of the use of antibiotics in animal feed, despite the growing opposition to antibiotics.
Unfortunately, the “charges” against the use of antibiotics like virginiamycin lacked a strong quantitative scientific basis. Phibro Animal Health, a leading producer of medical feed additives, needed objective models to quantify any human health risk posed by the use of virginiamycin in animal feed and to bring a neutral scientific perspective to the contentious issue for public policymakers.
The O.R. Solution
Phibro Animal Health retained Cox Associates, an independent O.R.-based risk analysis firm, to create such models and introduce specific numbers and probability estimates into policy and regulatory debates. The O.R. experts constructed models — applying Markov chain, Monte Carlo uncertainty analysis, and systems dynamics, using parameters that could be estimated and validated from publicly available historical data — to generate those numbers.
The models had to address the existence of certain basic uncertainties, including the fact that not all environmental pathways from animal microbial uses to resistant infections in humans, are known. The models had to identify effective and ineffective risk management interventions, including policy options between a simple virginiamycin ban on the one hand and a complete absence of regulation on the other. Finally, the O.R. models had to grapple with remaining uncertainties, and to assess the strength of the models’ conclusions and the value of performing additional scientific study to resolve residual unknowns on the basic health safety issues.
The O.R.-based analysis of virginiamycin in the United States revealed that the maximum harm that continued use might cause was less than a single “excess treatment failure” in the entire U.S. population over the next few years. In addition, the Markov chain and Monte Carlo modeling showed that the probability of Americans’ losing the ability to benefit from direct use of the human medical antibiotic equivalent to virginiamycin were essentially nonexistent.
The upshot of this analysis was a shift in the public health discussion from the risk of antibiotic resistance to minimizing human illness. That shift was possible once it became evident that the prudent use of antibiotics in feed is likely to prevent thousands of illnesses (by killing virginiamycin-susceptible bacteria in food animals). In addition, a new, more open-minded regulatory climate halted the decline in virginiamycin sales.