Antimicrobials are extensively used both prophylactically and therapeutically in poultry production. Despite this, there are little data on the effect of antimicrobial use (AMU) on disease incidence rate and per cent mortality. We investigated the relationships between AMU and disease and between AMU and mortality using data from a large (n = 322 flocks) cohort of small-scale chicken flocks in the Mekong Delta, Vietnam, that were followed longitudinally from day old to slaughter (5,566 observation weeks). We developed a parameterized algorithm to emulate a randomized control trial from observational data by categorizing the observation weeks into 'non-AMU', 'prophylactic AMU' and 'therapeutic AMU'. To evaluate the prophylactic AMU effect, we compared the frequencies of clinical signs in 'non-AMU' and 'prophylactic AMU' periods. To analyse therapeutic AMU, we compared weekly per cent mortality between the weeks of disease episodes before and after AMU. Analyses were stratified by clinical signs (4) and antimicrobial classes (13). Prophylactic AMU never reduced the probability of disease, and some antimicrobial classes such as lincosamides, amphenicols and penicillins increased the risk. The risk of diarrhoea consistently increased with prophylactic AMU. Therapeutic AMU often had an effect on mortality, but the pattern was inconsistent across the combinations of antimicrobial classes and clinical signs with 14/29 decreasing and 11/29 increasing the per cent weekly mortality. Lincosamides, methenamines and cephalosporins were the only three antimicrobial classes that always decreased the mortality when used therapeutically. Results were robust respective to the parameters values of the weeks categorization algorithm. This information should help support policy efforts and interventions aiming at reducing AMU in animal production.
Zoonoses and public health
483 - 492
Oxford University Clinical Research Unit, Ho Chi Minh City, Vietnam.