Letters to the Editor

Abstract

We read with interest the above-referenced article in the Journal of Food Protection (3). The issue of antimicrobial resistance in foodborne pathogens resulting from the use of antimicrobials in food-producing animals continues to be of concern to public health. We were pleased that the authors recognized this problem and that they were supportive of the U.S. Food and Drug Administration (FDA) guidance document (6) on the preapproval microbial safety assessment of antimicrobial agents intended for use in foodproducing animals. The authors’ quantitative risk assessment on macrolides (tylosin and tilmicosin) incorporated drug use data, which are critical for risk assessment but rarely available, at least in the United States. However, the model estimated drug use as the number of animals exposed rather than the actual amount of drug used. For example, tylosin given in feed for 60 days was counted as one exposure. A single injection of tilmicosin was also counted as one exposure. We know that long-term use of antimicrobial drugs exerts greater selection pressure than a one-time dose, so this parameter was inappropriately estimated (8). The hazard was defined in the article as human illness caused by macrolide-resistant Campylobacter species or Enterococcus faecium attributed to eating contaminated poultry or meat and treated with a human macrolide antimicrobial. This definition is appropriate for Campylobacter, but because the hazard as defined in the article ignores horizontal gene transfer, it is not sufficient for a risk assessment concerning enterococci. The authors agreed that enterococci are more important for their role in transferring resistance determinants than as direct causes of foodborne disease (3). Cross-contamination events also were not included in the model, although we know that contamination of animals and foods occurs during transport, lairage, slaughter, and processing and in the consumer’s kitchen during food preparation. The authors assumed that 1% of poultry treated with macrolides will harbor macrolide-resistant Campylobacter. This estimate is low. McDermott et al. (4) treated chickens with enrofloxacin, and 100% of the Campylobacter-infected chickens had fluoroquinolone-resistant Campylobacter versus 0% in the untreated group. Given the differences in the development and mechanisms of resistance between fluoroquinolones and macrolides, it is not reasonable to expect that 100% of poultry will harbor macrolide-resistant Campylobacter, but 10% is probably a more realistic estimate than 1%. Part of the low estimate used is based on the authors’ assumption that the amount of poultry meat contaminated with Campylobacter at the retail level is less than that measured on carcasses immediately after slaughter, citing a study published in 2001 (9). However, larger and more recent studies of retail meat contradict this assumption (2). For macrolide-resistant enterococci, the ratio between proportion of contaminated broilers and contaminated chicken meat samples is nearly 1:1 (1). The questionable assumptions applied tend to lower the estimated human exposure, resulting in overall low probability numbers calculated by Hurd et al. (3). Moreover, the probability of human illness is given per serving of poultry (or beef or pork), which does not equate to an annual risk. At a minimum, the numbers for poultry would need to be multiplied by 100 (conservatively assuming two poultry meals per week). Such a correction raises the risk substantially. For example, for the probability of human illness from macrolide-resistant Campylobacter in poultry, the risk rises from 1 in 14 million to 1 in 140,000, or approximately 2,000 people per year in the United States. With this simple conservative correction for number of servings of poultry eaten per year, the risk associated with macrolide use in poultry is of the same order of magnitude as that associated with enrofloxacin. The FDA Center for Veterinary Medicine is in the process of regulatory action to prohibit the use of enrofloxacin in poultry, in which a risk assessment on fluoroquinolone-resistant Campylobacter infections in humans arising from use of fluoroquinolones in chickens played a key role (5). The Administrative Law Judge recently issued an initial decision agreeing with the Center’s action and cited the above-mentioned risk assessment as demonstrating an unacceptable risk to human health (7). That decision is vastly different from the authors’ conclusion that the risk of macrolide-resistant Campylobacter species from tylosin and tilmicosin use in food-producing animals is ‘‘very low or remote.’’