The use of a task based exposure assessment approach during incident II resulted in 17 (24%) fewer workers categorised as being at risk and prescribed oseltamivir prophylaxis. This approach allowed the exclusion of a significant number of workers in the very low risk category D group (particularly long distance truck drivers and administrative staff) and limiting the number in high risk categories (A & B). Due to a communication error, seven category D workers did receive antiviral prophylaxis (but not PPE) at the onset of incident II. The increase in the proportion of category C workers receiving the interventions (including antiviral prophylaxis and PPE) during incident II is due to a precautionary increase in supervisory staff (duty and production managers) during protracted shift work. The reduction in workers requiring seasonal influenza vaccine during incident II is only in part due to task based exposure assessment, as some would have had prior immunisation.
The findings from an exposure risk assessment of carcass disposal options appears to suggest that exposure to and the risk of H1N1 infection is likely to be low in the latter stages of the carcass disposal process such as during the rendering of killed poultry.
However, specific guidance on risk assessment and management of rendering workers is not available and related guidance is conflicting. For example, the European Centre for Disease Prevention and Control guidance classifies rendering workers under the "direct contact" group . Occupational health and safety guidance in Northern Ireland excludes rendering workers from the at-risk group as they are judged not to have "direct contact with sick, dead or dying infected birds or bird products" . The existing Health Protection Agency guidance also does not specifically cover rendering workers and the application of the guidance to this population is open to interpretation . In addition, in the UK, rendering plant workers are specifically excluded from guidance recommending annual pre-exposure seasonal influenza immunisation of poultry workers .
Although the uniform risk assessment approach (using the "infected premises" principle) adopted in "Incident I" can be implemented rapidly and easily enforced, it lacks the specificity required when advising anxious workers on the need for antiviral prophylaxis. The uniform approach clearly has merits during culling operations on poultry farms with large numbers of temporary workers drafted in at short notice. By contrast, the controlled rendering plant environment is well suited to a task based exposure assessment approach due to the clear demarcation of work areas, operational processes and associated occupational functions. Adopting task based worker exposure assessment prior to HPAI outbreaks has application to all types of poultry industry workers as a means to ensuring a risk based and proportionate response to such situations.
The safety and efficacy of neuraminidase inhibitors has already been demonstrated extensively in studies of adults taking chemoprophylaxis following exposure to the seasonal influenza virus [21, 22]. However, there is a paucity of research evidence on the safety and efficacy of these agents against H5N1 particularly when used for short and longer-term prophylaxis particularly in low-risk occupational groups . Although we did not monitor compliance and measure the incidence of adverse events in our study population, it is possible that minor adverse events went undetected.
This study proposes improvements in the risk assessment methodology based on an understanding of the occupational and industrial process and was not designed to estimate the actual level of infection risk in rendering workers. The total absence of reported influenza like illness in workers up to 14 days after exposure to the rendering process, despite adopting a task based exposure assessment approach, is encouraging. Although it is possible that very mild cases may not have been detected, we have no reason to believe there was under-ascertainment of influenza like illness cases by general practitioners serving this workforce. A definitive evaluation of the potential for avian-to-human transmission of H5N1 in rendering workers would necessitate undertaking a cohort study comparing paired serum samples in all occupational categories over both incidents. Although there is evidence of potential for transmission in poultry workers and cullers, their exposure risk is significantly higher than rendering workers [13, 14]. This study also does not measure the added benefit of antiviral prophylaxis in comparison to barrier methods in preventing infection.
Even though the full workforce complement (by category and person-hours worked) at the plant remained largely unchanged, there were seasonal workers employed during the two incidents. This means that we cannot fully assume that the study population represents paired groups with observations taken from the same individuals during the two incidents, an assumption that is required to undertake appropriate statistical tests such as a McNemars test. We would have analysed the sub-set of the population with paired observations, but the heterogeneous nature of the study population (paired and unpaired) precluded our ability to link the observations recorded during the two incidents to specific individuals so as to identify discordant pairs.
The perception that the risk of infection and transmission is uniformly low in rendering workers has not been validated, particularly in terms of the higher risk occupational categories. Work to measure task based H5N1 exposure risk and biological outcomes (measures of infection and seroconversion) should be undertaken in future.